id	sid	tid	token	lemma	pos
work_55utqx7tjrft5ojtbr67ypjdye	1	1	Comparing	compare	VBG
work_55utqx7tjrft5ojtbr67ypjdye	1	2	Apples	apple	NNS
work_55utqx7tjrft5ojtbr67ypjdye	1	3	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	1	4	Apple	Apple	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	5	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	1	6	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	1	7	Effects	Effects	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	1	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	1	9	Stemmers	Stemmers	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	1	10	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	1	11	Topic	Topic	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	12	Models	Models	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	1	13	Alexandra	Alexandra	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	14	Schofield	Schofield	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	15	Cornell	Cornell	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	16	University	University	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	17	Ithaca	Ithaca	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	1	19	NY	NY	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	20	14853	14853	CD
work_55utqx7tjrft5ojtbr67ypjdye	1	21	xanda@cs.cornell.edu	xanda@cs.cornell.edu	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	22	David	David	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	23	Mimno	Mimno	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	24	Cornell	Cornell	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	25	University	University	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	26	Ithaca	Ithaca	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	27	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	1	28	NY	NY	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	29	14853	14853	CD
work_55utqx7tjrft5ojtbr67ypjdye	1	30	mimno@cornell.edu	mimno@cornell.edu	NN
work_55utqx7tjrft5ojtbr67ypjdye	1	31	Abstract	Abstract	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	32	Rule	Rule	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	33	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	1	34	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	1	35	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	1	36	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	1	37	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	1	38	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	1	39	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	40	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	1	41	mer	mer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	1	42	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	1	43	frequently	frequently	RB
work_55utqx7tjrft5ojtbr67ypjdye	1	44	used	use	VBN
work_55utqx7tjrft5ojtbr67ypjdye	1	45	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	1	46	preprocess	preprocess	VB
work_55utqx7tjrft5ojtbr67ypjdye	1	47	English	english	JJ
work_55utqx7tjrft5ojtbr67ypjdye	1	48	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	1	49	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	1	50	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	1	51	modeling	modeling	NN
work_55utqx7tjrft5ojtbr67ypjdye	1	52	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	2	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	2	2	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	2	3	work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	2	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	2	5	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	2	6	train	train	VBP
work_55utqx7tjrft5ojtbr67ypjdye	2	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	2	8	evaluate	evaluate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	2	9	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	2	10	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	2	11	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	2	12	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	2	13	variety	variety	NN
work_55utqx7tjrft5ojtbr67ypjdye	2	14	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	2	15	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	2	16	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	2	17	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	2	18	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	2	19	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	2	20	algo-	algo-	XX
work_55utqx7tjrft5ojtbr67ypjdye	2	21	rithms	rithms	NNP
work_55utqx7tjrft5ojtbr67ypjdye	2	22	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	3	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	3	2	examine	examine	VBP
work_55utqx7tjrft5ojtbr67ypjdye	3	3	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	3	4	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	3	5	quantita-	quantita-	NN
work_55utqx7tjrft5ojtbr67ypjdye	3	6	tive	tive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	3	7	measures	measure	NNS
work_55utqx7tjrft5ojtbr67ypjdye	3	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	3	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	3	10	resulting	result	VBG
work_55utqx7tjrft5ojtbr67ypjdye	3	11	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	3	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	3	13	includ-	includ-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	3	14	ing	ing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	3	15	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	3	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	3	17	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	3	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	3	19	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	3	20	stability	stability	NN
work_55utqx7tjrft5ojtbr67ypjdye	3	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	3	22	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	3	23	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	3	24	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	4	1	Despite	despite	IN
work_55utqx7tjrft5ojtbr67ypjdye	4	2	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	4	3	frequent	frequent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	4	4	use	use	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	5	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	4	6	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	7	modeling	modeling	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	4	9	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	4	10	find	find	VBP
work_55utqx7tjrft5ojtbr67ypjdye	4	11	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	4	12	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	4	13	produce	produce	VBP
work_55utqx7tjrft5ojtbr67ypjdye	4	14	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	4	15	meaningful	meaningful	JJ
work_55utqx7tjrft5ojtbr67ypjdye	4	16	improvement	improvement	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	17	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	4	18	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	19	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	4	20	co-	co-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	4	21	herence	herence	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	22	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	4	23	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	4	24	fact	fact	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	25	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	4	26	degrade	degrade	VB
work_55utqx7tjrft5ojtbr67ypjdye	4	27	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	28	stability	stability	NN
work_55utqx7tjrft5ojtbr67ypjdye	4	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	5	1	1	1	CD
work_55utqx7tjrft5ojtbr67ypjdye	5	2	Introduction	introduction	NN
work_55utqx7tjrft5ojtbr67ypjdye	5	3	Stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	5	4	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	5	5	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	5	6	popular	popular	JJ
work_55utqx7tjrft5ojtbr67ypjdye	5	7	way	way	NN
work_55utqx7tjrft5ojtbr67ypjdye	5	8	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	5	9	reduce	reduce	VB
work_55utqx7tjrft5ojtbr67ypjdye	5	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	5	11	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	5	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	5	13	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	5	14	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	5	15	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	5	16	natural	natural	JJ
work_55utqx7tjrft5ojtbr67ypjdye	5	17	language	language	NN
work_55utqx7tjrft5ojtbr67ypjdye	5	18	tasks	task	NNS
work_55utqx7tjrft5ojtbr67ypjdye	5	19	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	5	20	conflating	conflate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	5	21	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	5	22	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	5	23	related	related	JJ
work_55utqx7tjrft5ojtbr67ypjdye	5	24	meanings	meaning	NNS
work_55utqx7tjrft5ojtbr67ypjdye	5	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	6	1	Specifically	specifically	RB
work_55utqx7tjrft5ojtbr67ypjdye	6	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	6	3	stem-	stem-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	6	4	ming	ming	NNP
work_55utqx7tjrft5ojtbr67ypjdye	6	5	aims	aim	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	6	6	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	6	7	convert	convert	VB
work_55utqx7tjrft5ojtbr67ypjdye	6	8	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	6	9	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	6	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	6	11	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	6	12	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	6	13	stem	stem	NN
work_55utqx7tjrft5ojtbr67ypjdye	6	14	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	6	15	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	6	16	root	root	NN
work_55utqx7tjrft5ojtbr67ypjdye	6	17	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	6	18	e.g	e.g	NNP
work_55utqx7tjrft5ojtbr67ypjdye	6	19	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	6	20	creative	creative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	6	21	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	6	22	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	6	23	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	6	24	creator	creator	NN
work_55utqx7tjrft5ojtbr67ypjdye	6	25	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	6	26	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	6	27	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	6	28	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	6	29	single	single	JJ
work_55utqx7tjrft5ojtbr67ypjdye	6	30	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	6	31	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	6	32	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	6	33	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	6	34	create	create	VB
work_55utqx7tjrft5ojtbr67ypjdye	6	35	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	6	36	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	6	37	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	7	1	Though	though	IN
work_55utqx7tjrft5ojtbr67ypjdye	7	2	originally	originally	RB
work_55utqx7tjrft5ojtbr67ypjdye	7	3	developed	develop	VBN
work_55utqx7tjrft5ojtbr67ypjdye	7	4	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	7	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	7	6	context	context	NN
work_55utqx7tjrft5ojtbr67ypjdye	7	7	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	7	8	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	7	9	retrieval	retrieval	NN
work_55utqx7tjrft5ojtbr67ypjdye	7	10	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	7	11	IR	IR	NNP
work_55utqx7tjrft5ojtbr67ypjdye	7	12	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	7	13	systems	system	NNS
work_55utqx7tjrft5ojtbr67ypjdye	7	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	7	15	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	7	16	mers	mer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	7	17	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	7	18	now	now	RB
work_55utqx7tjrft5ojtbr67ypjdye	7	19	commonly	commonly	RB
work_55utqx7tjrft5ojtbr67ypjdye	7	20	used	use	VBN
work_55utqx7tjrft5ojtbr67ypjdye	7	21	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	7	22	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	7	23	preprocessing	preprocesse	VBG
work_55utqx7tjrft5ojtbr67ypjdye	7	24	step	step	NN
work_55utqx7tjrft5ojtbr67ypjdye	7	25	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	7	26	unsupervised	unsupervised	JJ
work_55utqx7tjrft5ojtbr67ypjdye	7	27	machine	machine	NN
work_55utqx7tjrft5ojtbr67ypjdye	7	28	learning	learning	NN
work_55utqx7tjrft5ojtbr67ypjdye	7	29	tasks	task	NNS
work_55utqx7tjrft5ojtbr67ypjdye	7	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	8	1	It	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	8	2	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	8	3	work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	8	4	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	8	5	consider	consider	VBP
work_55utqx7tjrft5ojtbr67ypjdye	8	6	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	8	7	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	8	8	application	application	NN
work_55utqx7tjrft5ojtbr67ypjdye	8	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	8	10	topic	topic	NNP
work_55utqx7tjrft5ojtbr67ypjdye	8	11	model-	model-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	8	12	ing	ing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	8	13	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	9	1	Although	although	IN
work_55utqx7tjrft5ojtbr67ypjdye	9	2	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	9	3	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	9	4	commonly	commonly	RB
work_55utqx7tjrft5ojtbr67ypjdye	9	5	used	use	VBN
work_55utqx7tjrft5ojtbr67ypjdye	9	6	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	9	7	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	9	8	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	9	9	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	9	10	Liu	Liu	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	11	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	12	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	13	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	9	15	2010	2010	CD
work_55utqx7tjrft5ojtbr67ypjdye	9	16	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	9	17	Lo	Lo	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	18	et	et	FW
work_55utqx7tjrft5ojtbr67ypjdye	9	19	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	20	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	9	22	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	9	23	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	9	24	Nan	Nan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	25	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	26	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	27	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	28	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	9	29	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	9	30	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	9	31	Kamath	Kamath	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	32	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	33	et	et	FW
work_55utqx7tjrft5ojtbr67ypjdye	9	34	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	35	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	36	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	9	37	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	9	38	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	9	39	Su	Su	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	40	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	9	41	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	9	42	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	9	43	Jacobi	Jacobi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	44	et	et	FW
work_55utqx7tjrft5ojtbr67ypjdye	9	45	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	46	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	9	47	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	9	48	2016	2016	CD
work_55utqx7tjrft5ojtbr67ypjdye	9	49	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	9	50	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	9	51	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	9	52	find	find	VBP
work_55utqx7tjrft5ojtbr67ypjdye	9	53	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	9	54	empirical	empirical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	9	55	benefits	benefit	NNS
work_55utqx7tjrft5ojtbr67ypjdye	9	56	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	9	57	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	9	58	practice	practice	NN
work_55utqx7tjrft5ojtbr67ypjdye	9	59	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	10	1	One	one	PRP
work_55utqx7tjrft5ojtbr67ypjdye	10	2	could	could	MD
work_55utqx7tjrft5ojtbr67ypjdye	10	3	conjecture	conjecture	VB
work_55utqx7tjrft5ojtbr67ypjdye	10	4	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	10	5	reasons	reason	NNS
work_55utqx7tjrft5ojtbr67ypjdye	10	6	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	10	7	stem	stem	VB
work_55utqx7tjrft5ojtbr67ypjdye	10	8	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	10	9	semantic	semantic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	10	10	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	10	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	11	1	First	first	RB
work_55utqx7tjrft5ojtbr67ypjdye	11	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	11	3	conflating	conflate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	11	4	semanti-	semanti-	NN
work_55utqx7tjrft5ojtbr67ypjdye	11	5	cally	cally	RB
work_55utqx7tjrft5ojtbr67ypjdye	11	6	related	relate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	11	7	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	11	8	into	into	IN
work_55utqx7tjrft5ojtbr67ypjdye	11	9	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	11	10	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	11	11	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	11	12	could	could	MD
work_55utqx7tjrft5ojtbr67ypjdye	11	13	im-	im-	RB
work_55utqx7tjrft5ojtbr67ypjdye	11	14	prove	prove	VB
work_55utqx7tjrft5ojtbr67ypjdye	11	15	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	11	16	fit	fit	JJ
work_55utqx7tjrft5ojtbr67ypjdye	11	17	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	11	18	intelligently	intelligently	RB
work_55utqx7tjrft5ojtbr67ypjdye	11	19	reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	11	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	11	21	space	space	NN
work_55utqx7tjrft5ojtbr67ypjdye	11	22	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	11	23	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	11	24	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	11	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	12	1	Given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	12	2	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	12	3	reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	12	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	12	5	fea-	fea-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	12	6	ture	ture	NN
work_55utqx7tjrft5ojtbr67ypjdye	12	7	space	space	NN
work_55utqx7tjrft5ojtbr67ypjdye	12	8	randomly	randomly	RB
work_55utqx7tjrft5ojtbr67ypjdye	12	9	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	12	10	already	already	RB
work_55utqx7tjrft5ojtbr67ypjdye	12	11	known	know	VBN
work_55utqx7tjrft5ojtbr67ypjdye	12	12	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	12	13	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	12	14	poten-	poten-	NN
work_55utqx7tjrft5ojtbr67ypjdye	12	15	tially	tially	RB
work_55utqx7tjrft5ojtbr67ypjdye	12	16	beneficial	beneficial	JJ
work_55utqx7tjrft5ojtbr67ypjdye	12	17	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	12	18	Ganchev	Ganchev	NNP
work_55utqx7tjrft5ojtbr67ypjdye	12	19	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	12	20	Dredze	Dredze	NNP
work_55utqx7tjrft5ojtbr67ypjdye	12	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	12	22	2008	2008	CD
work_55utqx7tjrft5ojtbr67ypjdye	12	23	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	12	24	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	12	25	do-	do-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	12	26	ing	e	VBG
work_55utqx7tjrft5ojtbr67ypjdye	12	27	so	so	RB
work_55utqx7tjrft5ojtbr67ypjdye	12	28	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	12	29	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	12	30	semantically	semantically	RB
work_55utqx7tjrft5ojtbr67ypjdye	12	31	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	12	32	inspired	inspire	VBN
work_55utqx7tjrft5ojtbr67ypjdye	12	33	way	way	NN
work_55utqx7tjrft5ojtbr67ypjdye	12	34	might	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	12	35	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	12	36	even	even	RB
work_55utqx7tjrft5ojtbr67ypjdye	12	37	better	well	JJR
work_55utqx7tjrft5ojtbr67ypjdye	12	38	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	13	1	Second	second	JJ
work_55utqx7tjrft5ojtbr67ypjdye	13	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	13	3	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	13	4	could	could	MD
work_55utqx7tjrft5ojtbr67ypjdye	13	5	reduce	reduce	VB
work_55utqx7tjrft5ojtbr67ypjdye	13	6	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	13	7	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	13	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	13	9	small	small	JJ
work_55utqx7tjrft5ojtbr67ypjdye	13	10	morphological	morphological	JJ
work_55utqx7tjrft5ojtbr67ypjdye	13	11	differences	difference	NNS
work_55utqx7tjrft5ojtbr67ypjdye	13	12	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	13	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	13	14	stability	stability	NN
work_55utqx7tjrft5ojtbr67ypjdye	13	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	13	16	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	13	17	learned	learn	VBN
work_55utqx7tjrft5ojtbr67ypjdye	13	18	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	13	19	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	14	1	Reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	14	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	14	3	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	14	4	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	14	5	happy	happy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	14	6	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	14	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	14	8	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	14	9	happily	happily	RB
work_55utqx7tjrft5ojtbr67ypjdye	14	10	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	14	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	14	12	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	14	13	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	14	14	happier	happy	JJR
work_55utqx7tjrft5ojtbr67ypjdye	14	15	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	14	16	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	14	17	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	14	18	token	token	NN
work_55utqx7tjrft5ojtbr67ypjdye	14	19	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	14	20	result	result	VB
work_55utqx7tjrft5ojtbr67ypjdye	14	21	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	14	22	fewer	few	JJR
work_55utqx7tjrft5ojtbr67ypjdye	14	23	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	14	24	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	14	25	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	14	26	divergent	divergent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	14	27	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	14	28	happy	happy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	14	29	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	14	30	top-	top-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	14	31	ics	ics	NNP
work_55utqx7tjrft5ojtbr67ypjdye	14	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	15	1	Third	third	JJ
work_55utqx7tjrft5ojtbr67ypjdye	15	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	15	3	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	15	4	approximate	approximate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	15	5	intuitive	intuitive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	15	6	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	15	7	equivalence	equivalence	NN
work_55utqx7tjrft5ojtbr67ypjdye	15	8	classes	class	NNS
work_55utqx7tjrft5ojtbr67ypjdye	15	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	15	10	so	so	CC
work_55utqx7tjrft5ojtbr67ypjdye	15	11	language	language	NN
work_55utqx7tjrft5ojtbr67ypjdye	15	12	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	15	13	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	15	14	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	15	15	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	15	16	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	15	17	inherit	inherit	NN
work_55utqx7tjrft5ojtbr67ypjdye	15	18	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	15	19	semantic	semantic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	15	20	similarity	similarity	NN
work_55utqx7tjrft5ojtbr67ypjdye	15	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	15	22	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	15	23	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	15	24	improve	improve	VB
work_55utqx7tjrft5ojtbr67ypjdye	15	25	interpretability	interpretability	NN
work_55utqx7tjrft5ojtbr67ypjdye	15	26	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	15	27	perceived	perceive	VBN
work_55utqx7tjrft5ojtbr67ypjdye	15	28	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	15	29	human	human	JJ
work_55utqx7tjrft5ojtbr67ypjdye	15	30	evaluators	evaluator	NNS
work_55utqx7tjrft5ojtbr67ypjdye	15	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	16	1	However	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	16	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	16	3	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	16	4	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	16	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	16	6	potential	potential	NN
work_55utqx7tjrft5ojtbr67ypjdye	16	7	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	16	8	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	16	9	con-	con-	NN
work_55utqx7tjrft5ojtbr67ypjdye	16	10	fusing	fuse	VBG
work_55utqx7tjrft5ojtbr67ypjdye	16	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	16	12	unreliable	unreliable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	16	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	16	14	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	16	15	possibly	possibly	RB
work_55utqx7tjrft5ojtbr67ypjdye	16	16	even	even	RB
work_55utqx7tjrft5ojtbr67ypjdye	16	17	harmful	harmful	JJ
work_55utqx7tjrft5ojtbr67ypjdye	16	18	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	16	19	lan-	lan-	NN
work_55utqx7tjrft5ojtbr67ypjdye	16	20	guage	guage	NN
work_55utqx7tjrft5ojtbr67ypjdye	16	21	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	16	22	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	17	1	First	first	RB
work_55utqx7tjrft5ojtbr67ypjdye	17	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	17	3	many	many	JJ
work_55utqx7tjrft5ojtbr67ypjdye	17	4	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	17	5	produce	produce	VBP
work_55utqx7tjrft5ojtbr67ypjdye	17	6	terms	term	NNS
work_55utqx7tjrft5ojtbr67ypjdye	17	7	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	17	8	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	17	9	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	17	10	recognizable	recognizable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	17	11	English	english	JJ
work_55utqx7tjrft5ojtbr67ypjdye	17	12	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	17	13	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	17	14	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	17	15	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	17	16	difficult	difficult	JJ
work_55utqx7tjrft5ojtbr67ypjdye	17	17	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	17	18	map	map	VB
work_55utqx7tjrft5ojtbr67ypjdye	17	19	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	17	20	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	17	21	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	17	22	valid	valid	JJ
work_55utqx7tjrft5ojtbr67ypjdye	17	23	original	original	JJ
work_55utqx7tjrft5ojtbr67ypjdye	17	24	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	17	25	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	17	26	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	17	27	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	17	28	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	17	29	stai	stai	NNP
work_55utqx7tjrft5ojtbr67ypjdye	17	30	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	17	31	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	17	32	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	17	33	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	17	34	stem	stem	NN
work_55utqx7tjrft5ojtbr67ypjdye	17	35	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	17	36	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	17	37	stay	stay	VB
work_55utqx7tjrft5ojtbr67ypjdye	17	38	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	17	39	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	18	1	Sec-	Sec-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	18	2	ond	ond	NN
work_55utqx7tjrft5ojtbr67ypjdye	18	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	18	4	although	although	IN
work_55utqx7tjrft5ojtbr67ypjdye	18	5	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	18	6	aids	aids	NNP
work_55utqx7tjrft5ojtbr67ypjdye	18	7	document	document	NN
work_55utqx7tjrft5ojtbr67ypjdye	18	8	retrieval	retrieval	NN
work_55utqx7tjrft5ojtbr67ypjdye	18	9	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	18	10	many	many	JJ
work_55utqx7tjrft5ojtbr67ypjdye	18	11	languages	language	NNS
work_55utqx7tjrft5ojtbr67ypjdye	18	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	18	13	English	English	NNP
work_55utqx7tjrft5ojtbr67ypjdye	18	14	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	18	15	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	18	16	notorious	notorious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	18	17	exception	exception	NN
work_55utqx7tjrft5ojtbr67ypjdye	18	18	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	18	19	Harman	Harman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	18	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	18	21	1991	1991	CD
work_55utqx7tjrft5ojtbr67ypjdye	18	22	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	18	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	19	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	2	English	English	NNP
work_55utqx7tjrft5ojtbr67ypjdye	19	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	19	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	19	5	complexity	complexity	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	7	compound	compound	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	8	affixes	affix	NNS
work_55utqx7tjrft5ojtbr67ypjdye	19	9	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	10	meaning	meaning	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	11	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	19	12	lead	lead	VB
work_55utqx7tjrft5ojtbr67ypjdye	19	13	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	14	over-	over-	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	15	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	19	17	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	19	18	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	19	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	19	20	recondition	recondition	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	19	22	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	19	23	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	19	24	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	25	sharing	share	VBG
work_55utqx7tjrft5ojtbr67ypjdye	19	26	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	19	27	stem	stem	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	28	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	19	29	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	19	30	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	19	31	root	root	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	32	meaning	meaning	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	33	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	34	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	19	35	recondite	recondite	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	36	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	19	37	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	19	38	These	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	19	39	complexities	complexity	NNS
work_55utqx7tjrft5ojtbr67ypjdye	19	40	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	19	41	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	19	42	lead	lead	VB
work_55utqx7tjrft5ojtbr67ypjdye	19	43	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	44	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	19	45	incorrect	incorrect	JJ
work_55utqx7tjrft5ojtbr67ypjdye	19	46	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	47	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	48	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	19	49	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	50	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	19	51	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	19	52	root	root	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	53	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	19	54	divergent	divergent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	19	55	meaning	meaning	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	56	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	19	57	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	19	58	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	19	59	absolutely	absolutely	RB
work_55utqx7tjrft5ojtbr67ypjdye	19	60	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	19	61	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	19	62	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	19	63	absolution	absolution	NN
work_55utqx7tjrft5ojtbr67ypjdye	19	64	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	19	65	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	20	1	Third	third	JJ
work_55utqx7tjrft5ojtbr67ypjdye	20	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	20	3	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	20	4	most	most	RBS
work_55utqx7tjrft5ojtbr67ypjdye	20	5	troubling	troubling	JJ
work_55utqx7tjrft5ojtbr67ypjdye	20	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	20	7	there	there	EX
work_55utqx7tjrft5ojtbr67ypjdye	20	8	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	20	9	cases	case	NNS
work_55utqx7tjrft5ojtbr67ypjdye	20	10	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	20	11	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	20	12	morpholog-	morpholog-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	20	13	ical	ical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	20	14	variants	variant	NNS
work_55utqx7tjrft5ojtbr67ypjdye	20	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	20	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	20	17	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	20	18	stem	stem	NN
work_55utqx7tjrft5ojtbr67ypjdye	20	19	carry	carry	VBP
work_55utqx7tjrft5ojtbr67ypjdye	20	20	significantly	significantly	RB
work_55utqx7tjrft5ojtbr67ypjdye	20	21	dif-	dif-	CC
work_55utqx7tjrft5ojtbr67ypjdye	20	22	ferent	ferent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	20	23	meanings	meaning	NNS
work_55utqx7tjrft5ojtbr67ypjdye	20	24	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	21	1	Conflating	conflate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	21	2	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	21	3	apple	apple	NN
work_55utqx7tjrft5ojtbr67ypjdye	21	4	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	21	5	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	21	6	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	21	7	apples	apple	NNS
work_55utqx7tjrft5ojtbr67ypjdye	21	8	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	21	9	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	21	10	uncontroversial	uncontroversial	JJ
work_55utqx7tjrft5ojtbr67ypjdye	21	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	21	12	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	21	13	loses	lose	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	21	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	21	15	distinction	distinction	NN
work_55utqx7tjrft5ojtbr67ypjdye	21	16	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	21	17	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	21	18	device	device	NN
work_55utqx7tjrft5ojtbr67ypjdye	21	19	manufacturer	manufacturer	NN
work_55utqx7tjrft5ojtbr67ypjdye	21	20	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	21	21	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	21	22	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	21	23	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	21	24	fruit	fruit	NN
work_55utqx7tjrft5ojtbr67ypjdye	21	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	22	1	287	287	CD
work_55utqx7tjrft5ojtbr67ypjdye	22	2	Transactions	transaction	NNS
work_55utqx7tjrft5ojtbr67ypjdye	22	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	22	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	22	5	Association	Association	NNP
work_55utqx7tjrft5ojtbr67ypjdye	22	6	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	22	7	Computational	Computational	NNP
work_55utqx7tjrft5ojtbr67ypjdye	22	8	Linguistics	Linguistics	NNP
work_55utqx7tjrft5ojtbr67ypjdye	22	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	22	10	vol	vol	NNP
work_55utqx7tjrft5ojtbr67ypjdye	22	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	23	1	4	4	LS
work_55utqx7tjrft5ojtbr67ypjdye	23	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	23	3	pp	pp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	23	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	24	1	287–300	287–300	CD
work_55utqx7tjrft5ojtbr67ypjdye	24	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	24	3	2016	2016	CD
work_55utqx7tjrft5ojtbr67ypjdye	24	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	25	1	Action	Action	NNP
work_55utqx7tjrft5ojtbr67ypjdye	25	2	Editor	Editor	NNP
work_55utqx7tjrft5ojtbr67ypjdye	25	3	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	25	4	Hal	Hal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	25	5	Daume	Daume	NNP
work_55utqx7tjrft5ojtbr67ypjdye	25	6	III	III	NNP
work_55utqx7tjrft5ojtbr67ypjdye	25	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	26	1	Submission	submission	NN
work_55utqx7tjrft5ojtbr67ypjdye	26	2	batch	batch	NN
work_55utqx7tjrft5ojtbr67ypjdye	26	3	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	26	4	2/2016	2/2016	CD
work_55utqx7tjrft5ojtbr67ypjdye	26	5	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	26	6	Published	publish	VBN
work_55utqx7tjrft5ojtbr67ypjdye	26	7	7/2016	7/2016	CD
work_55utqx7tjrft5ojtbr67ypjdye	26	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	27	1	c	c	NNP
work_55utqx7tjrft5ojtbr67ypjdye	27	2	©	©	NNP
work_55utqx7tjrft5ojtbr67ypjdye	27	3	2016	2016	CD
work_55utqx7tjrft5ojtbr67ypjdye	27	4	Association	Association	NNP
work_55utqx7tjrft5ojtbr67ypjdye	27	5	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	27	6	Computational	Computational	NNP
work_55utqx7tjrft5ojtbr67ypjdye	27	7	Linguistics	Linguistics	NNP
work_55utqx7tjrft5ojtbr67ypjdye	27	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	28	1	Distributed	distribute	VBN
work_55utqx7tjrft5ojtbr67ypjdye	28	2	under	under	IN
work_55utqx7tjrft5ojtbr67ypjdye	28	3	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	28	4	CC	cc	NN
work_55utqx7tjrft5ojtbr67ypjdye	28	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	28	6	BY	BY	NNP
work_55utqx7tjrft5ojtbr67ypjdye	28	7	4.0	4.0	CD
work_55utqx7tjrft5ojtbr67ypjdye	28	8	license	license	NN
work_55utqx7tjrft5ojtbr67ypjdye	28	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	29	1	Topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	29	2	modeling	modeling	NN
work_55utqx7tjrft5ojtbr67ypjdye	29	3	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	29	4	sensitive	sensitive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	29	5	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	29	6	preprocessing	preprocesse	VBG
work_55utqx7tjrft5ojtbr67ypjdye	29	7	be-	be-	XX
work_55utqx7tjrft5ojtbr67ypjdye	29	8	cause	cause	NN
work_55utqx7tjrft5ojtbr67ypjdye	29	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	29	10	its	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	29	11	dependence	dependence	NN
work_55utqx7tjrft5ojtbr67ypjdye	29	12	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	29	13	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	29	14	sparse	sparse	JJ
work_55utqx7tjrft5ojtbr67ypjdye	29	15	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	29	16	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	29	17	Jockers	Jockers	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	29	18	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	29	19	Mimno	Mimno	NNP
work_55utqx7tjrft5ojtbr67ypjdye	29	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	29	21	2013	2013	CD
work_55utqx7tjrft5ojtbr67ypjdye	29	22	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	29	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	30	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	30	2	practice	practice	NN
work_55utqx7tjrft5ojtbr67ypjdye	30	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	30	4	however	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	30	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	30	6	preprocessing	preprocesse	VBG
work_55utqx7tjrft5ojtbr67ypjdye	30	7	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	30	8	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	30	9	typically	typically	RB
work_55utqx7tjrft5ojtbr67ypjdye	30	10	neither	neither	CC
work_55utqx7tjrft5ojtbr67ypjdye	30	11	detailed	detail	VBN
work_55utqx7tjrft5ojtbr67ypjdye	30	12	nor	nor	CC
work_55utqx7tjrft5ojtbr67ypjdye	30	13	justified	justified	JJ
work_55utqx7tjrft5ojtbr67ypjdye	30	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	30	15	leading	lead	VBG
work_55utqx7tjrft5ojtbr67ypjdye	30	16	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	30	17	problems	problem	NNS
work_55utqx7tjrft5ojtbr67ypjdye	30	18	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	30	19	reproducibility	reproducibility	NN
work_55utqx7tjrft5ojtbr67ypjdye	30	20	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	30	21	Fokkens	Fokkens	NNP
work_55utqx7tjrft5ojtbr67ypjdye	30	22	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	30	23	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	30	24	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	30	25	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	30	26	2013	2013	CD
work_55utqx7tjrft5ojtbr67ypjdye	30	27	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	30	28	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	31	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	31	2	believe	believe	VBP
work_55utqx7tjrft5ojtbr67ypjdye	31	3	investigating	investigate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	31	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	31	5	effects	effect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	31	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	31	7	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	31	8	will	will	MD
work_55utqx7tjrft5ojtbr67ypjdye	31	9	inform	inform	VB
work_55utqx7tjrft5ojtbr67ypjdye	31	10	researchers	researcher	NNS
work_55utqx7tjrft5ojtbr67ypjdye	31	11	outside	outside	IN
work_55utqx7tjrft5ojtbr67ypjdye	31	12	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	31	13	core	core	JJ
work_55utqx7tjrft5ojtbr67ypjdye	31	14	natural	natural	JJ
work_55utqx7tjrft5ojtbr67ypjdye	31	15	language	language	NN
work_55utqx7tjrft5ojtbr67ypjdye	31	16	processing	processing	NN
work_55utqx7tjrft5ojtbr67ypjdye	31	17	community	community	NN
work_55utqx7tjrft5ojtbr67ypjdye	31	18	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	31	19	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	31	20	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	31	21	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	31	22	best	good	JJS
work_55utqx7tjrft5ojtbr67ypjdye	31	23	preprocess	preprocess	VB
work_55utqx7tjrft5ojtbr67ypjdye	31	24	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	31	25	texts	text	NNS
work_55utqx7tjrft5ojtbr67ypjdye	31	26	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	32	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	32	2	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	32	3	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	32	4	used	use	VBN
work_55utqx7tjrft5ojtbr67ypjdye	32	5	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	32	6	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	32	7	modeling	modeling	NN
work_55utqx7tjrft5ojtbr67ypjdye	32	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	32	9	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	32	10	know	know	VBP
work_55utqx7tjrft5ojtbr67ypjdye	32	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	32	12	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	32	13	analysis	analysis	NN
work_55utqx7tjrft5ojtbr67ypjdye	32	14	focused	focus	VBN
work_55utqx7tjrft5ojtbr67ypjdye	32	15	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	32	16	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	32	17	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	32	18	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	33	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	33	2	draw	draw	VBP
work_55utqx7tjrft5ojtbr67ypjdye	33	3	inspiration	inspiration	NN
work_55utqx7tjrft5ojtbr67ypjdye	33	4	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	33	5	prior	prior	JJ
work_55utqx7tjrft5ojtbr67ypjdye	33	6	studies	study	NNS
work_55utqx7tjrft5ojtbr67ypjdye	33	7	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	33	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	33	9	effects	effect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	33	10	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	33	11	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	33	12	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	33	13	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	33	14	tasks	task	NNS
work_55utqx7tjrft5ojtbr67ypjdye	33	15	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	33	16	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	33	17	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	33	18	Harman	Harman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	33	20	1991	1991	CD
work_55utqx7tjrft5ojtbr67ypjdye	33	21	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	33	22	Han	Han	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	23	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	24	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	25	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	26	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	33	27	2012	2012	CD
work_55utqx7tjrft5ojtbr67ypjdye	33	28	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	33	29	Jivani	Jivani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	30	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	33	31	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	33	32	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	33	33	Rani	Rani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	34	et	et	FW
work_55utqx7tjrft5ojtbr67ypjdye	33	35	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	36	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	33	37	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	33	38	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	33	39	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	33	40	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	33	41	apply	apply	VB
work_55utqx7tjrft5ojtbr67ypjdye	33	42	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	33	43	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	33	44	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	33	45	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	33	46	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	33	47	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	33	48	variety	variety	NN
work_55utqx7tjrft5ojtbr67ypjdye	33	49	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	33	50	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	33	51	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	33	52	test	test	VB
work_55utqx7tjrft5ojtbr67ypjdye	33	53	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	33	54	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	33	55	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	33	56	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	33	57	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	33	58	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	34	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	34	2	evaluate	evaluate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	34	3	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	34	4	quantitative	quantitative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	34	5	fit	fit	NN
work_55utqx7tjrft5ojtbr67ypjdye	34	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	34	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	34	8	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	34	9	generated	generate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	34	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	34	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	34	12	qualitative	qualitative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	34	13	differences	difference	NNS
work_55utqx7tjrft5ojtbr67ypjdye	34	14	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	34	15	differently-	differently-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	34	16	stemmed	stem	VBD
work_55utqx7tjrft5ojtbr67ypjdye	34	17	corpora	corpora	NNP
work_55utqx7tjrft5ojtbr67ypjdye	34	18	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	34	19	investigate	investigate	VB
work_55utqx7tjrft5ojtbr67ypjdye	34	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	34	21	effects	effect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	34	22	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	34	23	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	34	24	has	have	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	34	25	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	34	26	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	34	27	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	34	28	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	35	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	35	2	hope	hope	VBP
work_55utqx7tjrft5ojtbr67ypjdye	35	3	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	35	4	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	35	5	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	35	6	help	help	VBP
work_55utqx7tjrft5ojtbr67ypjdye	35	7	guide	guide	VB
work_55utqx7tjrft5ojtbr67ypjdye	35	8	future	future	JJ
work_55utqx7tjrft5ojtbr67ypjdye	35	9	researchers	researcher	NNS
work_55utqx7tjrft5ojtbr67ypjdye	35	10	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	35	11	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	35	12	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	35	13	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	35	14	select	select	VB
work_55utqx7tjrft5ojtbr67ypjdye	35	15	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	35	16	evaluate	evaluate	VB
work_55utqx7tjrft5ojtbr67ypjdye	35	17	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	35	18	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	35	19	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	35	20	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	35	21	task	task	NN
work_55utqx7tjrft5ojtbr67ypjdye	35	22	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	35	23	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	35	24	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	36	1	2	2	LS
work_55utqx7tjrft5ojtbr67ypjdye	36	2	Background	Background	NNP
work_55utqx7tjrft5ojtbr67ypjdye	36	3	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	36	4	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	36	5	work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	6	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	36	7	consider	consider	VBP
work_55utqx7tjrft5ojtbr67ypjdye	36	8	two	two	CD
work_55utqx7tjrft5ojtbr67ypjdye	36	9	categories	category	NNS
work_55utqx7tjrft5ojtbr67ypjdye	36	10	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	36	11	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	12	nor-	nor-	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	13	malization1	malization1	NNP
work_55utqx7tjrft5ojtbr67ypjdye	36	14	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	36	15	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	36	16	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	17	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	36	18	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	36	19	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	36	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	36	21	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	36	22	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	23	mers	mer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	36	24	primarily	primarily	RB
work_55utqx7tjrft5ojtbr67ypjdye	36	25	reliant	reliant	JJ
work_55utqx7tjrft5ojtbr67ypjdye	36	26	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	36	27	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	36	28	converting	convert	VBG
work_55utqx7tjrft5ojtbr67ypjdye	36	29	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	36	30	af-	af-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	36	31	fix	fix	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	32	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	36	33	another	another	DT
work_55utqx7tjrft5ojtbr67ypjdye	36	34	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	36	35	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	36	36	context	context	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	37	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	36	38	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	36	39	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	36	40	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	36	41	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	36	42	strate-	strate-	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	43	gies	gy	NNS
work_55utqx7tjrft5ojtbr67ypjdye	36	44	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	36	45	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	36	46	dictionaries	dictionary	NNS
work_55utqx7tjrft5ojtbr67ypjdye	36	47	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	36	48	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	36	49	contextual	contextual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	36	50	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	36	51	in-	in-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	36	52	flectional	flectional	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	53	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	36	54	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	36	55	derivational	derivational	JJ
work_55utqx7tjrft5ojtbr67ypjdye	36	56	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	57	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	36	58	infer	infer	VB
work_55utqx7tjrft5ojtbr67ypjdye	36	59	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	36	60	correct	correct	JJ
work_55utqx7tjrft5ojtbr67ypjdye	36	61	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	62	root	root	NN
work_55utqx7tjrft5ojtbr67ypjdye	36	63	Jivani	Jivani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	36	64	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	36	65	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	36	66	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	36	67	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	37	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	37	2	omit	omit	VBP
work_55utqx7tjrft5ojtbr67ypjdye	37	3	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	37	4	language	language	NN
work_55utqx7tjrft5ojtbr67ypjdye	37	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	37	6	independent	independent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	37	7	strategies	strategy	NNS
work_55utqx7tjrft5ojtbr67ypjdye	37	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	37	9	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	37	10	normaliza-	normaliza-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	37	11	tion	tion	NN
work_55utqx7tjrft5ojtbr67ypjdye	37	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	37	13	including	include	VBG
work_55utqx7tjrft5ojtbr67ypjdye	37	14	those	those	DT
work_55utqx7tjrft5ojtbr67ypjdye	37	15	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	37	16	Markov	Markov	NNP
work_55utqx7tjrft5ojtbr67ypjdye	37	17	chains	chain	NNS
work_55utqx7tjrft5ojtbr67ypjdye	37	18	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	37	19	Melucci	Melucci	NNP
work_55utqx7tjrft5ojtbr67ypjdye	37	20	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	37	21	Orio	Orio	NNP
work_55utqx7tjrft5ojtbr67ypjdye	37	22	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	37	23	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	37	24	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	37	25	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	37	26	clustering	cluster	VBG
work_55utqx7tjrft5ojtbr67ypjdye	37	27	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	37	28	Majumder	Majumder	NNP
work_55utqx7tjrft5ojtbr67ypjdye	37	29	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	37	30	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	37	31	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	37	32	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	37	33	2007	2007	CD
work_55utqx7tjrft5ojtbr67ypjdye	37	34	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	37	35	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	38	1	These	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	38	2	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	38	3	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	38	4	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	38	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	38	6	specific	specific	JJ
work_55utqx7tjrft5ojtbr67ypjdye	38	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	38	8	error-	error-	XX
work_55utqx7tjrft5ojtbr67ypjdye	38	9	prone	prone	JJ
work_55utqx7tjrft5ojtbr67ypjdye	38	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	38	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	38	12	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	38	13	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	38	14	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	38	15	observed	observe	VBN
work_55utqx7tjrft5ojtbr67ypjdye	38	16	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	38	17	use	use	NN
work_55utqx7tjrft5ojtbr67ypjdye	38	18	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	38	19	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	38	20	modeling	modeling	NN
work_55utqx7tjrft5ojtbr67ypjdye	38	21	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	39	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	39	2	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	39	3	evaluation	evaluation	NN
work_55utqx7tjrft5ojtbr67ypjdye	39	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	39	5	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	39	6	consider	consider	VBP
work_55utqx7tjrft5ojtbr67ypjdye	39	7	nine	nine	CD
work_55utqx7tjrft5ojtbr67ypjdye	39	8	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	39	9	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	39	10	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	39	11	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	39	12	normalization	normalization	NN
work_55utqx7tjrft5ojtbr67ypjdye	39	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	39	14	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	39	15	below	below	RP
work_55utqx7tjrft5ojtbr67ypjdye	39	16	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	39	17	two	two	CD
work_55utqx7tjrft5ojtbr67ypjdye	39	18	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	39	19	letter	letter	NN
work_55utqx7tjrft5ojtbr67ypjdye	39	20	labels	label	NNS
work_55utqx7tjrft5ojtbr67ypjdye	39	21	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	40	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	40	2	addition	addition	NN
work_55utqx7tjrft5ojtbr67ypjdye	40	3	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	40	4	including	include	VBG
work_55utqx7tjrft5ojtbr67ypjdye	40	5	popu-	popu-	XX
work_55utqx7tjrft5ojtbr67ypjdye	40	6	lar	lar	NNP
work_55utqx7tjrft5ojtbr67ypjdye	40	7	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	40	8	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	40	9	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	40	10	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	40	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	40	12	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	40	13	choose	choose	VBP
work_55utqx7tjrft5ojtbr67ypjdye	40	14	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	40	15	sim-	sim-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	40	16	ple	ple	NN
work_55utqx7tjrft5ojtbr67ypjdye	40	17	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	40	18	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	40	19	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	40	20	stronger	strong	JJR
work_55utqx7tjrft5ojtbr67ypjdye	40	21	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	40	22	weaker	weak	JJR
work_55utqx7tjrft5ojtbr67ypjdye	40	23	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	40	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	40	25	named	name	VBN
work_55utqx7tjrft5ojtbr67ypjdye	40	26	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	40	27	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	40	28	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	40	29	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	40	30	refers	refer	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	40	31	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	40	32	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	40	33	much	much	JJ
work_55utqx7tjrft5ojtbr67ypjdye	40	34	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	40	35	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	40	36	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	40	37	reduced	reduce	VBN
work_55utqx7tjrft5ojtbr67ypjdye	40	38	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	41	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	41	2	will	will	MD
work_55utqx7tjrft5ojtbr67ypjdye	41	3	sometimes	sometimes	RB
work_55utqx7tjrft5ojtbr67ypjdye	41	4	use	use	VB
work_55utqx7tjrft5ojtbr67ypjdye	41	5	1Other	1other	CD
work_55utqx7tjrft5ojtbr67ypjdye	41	6	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	41	7	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	41	8	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	41	9	normalization	normalization	NN
work_55utqx7tjrft5ojtbr67ypjdye	41	10	include	include	VBP
work_55utqx7tjrft5ojtbr67ypjdye	41	11	case	case	NN
work_55utqx7tjrft5ojtbr67ypjdye	41	12	folding	fold	VBG
work_55utqx7tjrft5ojtbr67ypjdye	41	13	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	41	14	replacing	replace	VBG
work_55utqx7tjrft5ojtbr67ypjdye	41	15	classes	class	NNS
work_55utqx7tjrft5ojtbr67ypjdye	41	16	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	41	17	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	41	18	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	41	19	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	41	20	constant	constant	NN
work_55utqx7tjrft5ojtbr67ypjdye	41	21	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	41	22	e.g.	e.g.	RB
work_55utqx7tjrft5ojtbr67ypjdye	42	1	NUMBER	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	42	2	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	42	3	numerals	numeral	NNS
work_55utqx7tjrft5ojtbr67ypjdye	42	4	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	42	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	43	1	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	43	2	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	43	3	general	general	JJ
work_55utqx7tjrft5ojtbr67ypjdye	43	4	term	term	NN
work_55utqx7tjrft5ojtbr67ypjdye	43	5	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	43	6	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	43	7	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	43	8	sim-	sim-	NN
work_55utqx7tjrft5ojtbr67ypjdye	43	9	ply	ply	UH
work_55utqx7tjrft5ojtbr67ypjdye	43	10	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	43	11	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	43	12	refer	refer	VB
work_55utqx7tjrft5ojtbr67ypjdye	43	13	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	43	14	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	43	15	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	43	16	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	43	17	respect	respect	NN
work_55utqx7tjrft5ojtbr67ypjdye	43	18	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	43	19	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	43	20	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	43	21	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	44	1	These	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	44	2	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	44	3	compared	compare	VBN
work_55utqx7tjrft5ojtbr67ypjdye	44	4	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	44	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	44	6	control	control	NN
work_55utqx7tjrft5ojtbr67ypjdye	44	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	44	8	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	44	9	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	44	10	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	44	11	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	44	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	44	13	NS	NS	NNP
work_55utqx7tjrft5ojtbr67ypjdye	44	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	45	1	2.1	2.1	CD
work_55utqx7tjrft5ojtbr67ypjdye	45	2	Rule	Rule	NNP
work_55utqx7tjrft5ojtbr67ypjdye	45	3	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	45	4	Based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	45	5	Treatments	Treatments	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	45	6	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	45	7	first	first	JJ
work_55utqx7tjrft5ojtbr67ypjdye	45	8	category	category	NN
work_55utqx7tjrft5ojtbr67ypjdye	45	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	45	10	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	45	11	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	45	12	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	45	13	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	45	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	45	15	includes	include	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	45	16	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	45	17	primarily	primarily	RB
work_55utqx7tjrft5ojtbr67ypjdye	45	18	governed	govern	VBN
work_55utqx7tjrft5ojtbr67ypjdye	45	19	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	45	20	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	45	21	set	set	NN
work_55utqx7tjrft5ojtbr67ypjdye	45	22	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	45	23	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	45	24	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	45	25	convert	convert	VBP
work_55utqx7tjrft5ojtbr67ypjdye	45	26	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	45	27	affix	affix	NN
work_55utqx7tjrft5ojtbr67ypjdye	45	28	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	45	29	another	another	DT
work_55utqx7tjrft5ojtbr67ypjdye	45	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	46	1	Most	Most	JJS
work_55utqx7tjrft5ojtbr67ypjdye	46	2	classic	classic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	46	3	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	46	4	fit	fit	VBP
work_55utqx7tjrft5ojtbr67ypjdye	46	5	into	into	IN
work_55utqx7tjrft5ojtbr67ypjdye	46	6	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	46	7	category	category	NN
work_55utqx7tjrft5ojtbr67ypjdye	46	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	46	9	including	include	VBG
work_55utqx7tjrft5ojtbr67ypjdye	46	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	46	11	famous	famous	JJ
work_55utqx7tjrft5ojtbr67ypjdye	46	12	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	46	13	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	46	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	47	1	These	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	47	2	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	47	3	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	47	4	quick	quick	JJ
work_55utqx7tjrft5ojtbr67ypjdye	47	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	47	6	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	47	7	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	47	8	limited	limited	JJ
work_55utqx7tjrft5ojtbr67ypjdye	47	9	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	47	10	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	47	11	concise	concise	JJ
work_55utqx7tjrft5ojtbr67ypjdye	47	12	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	47	13	set	set	VBD
work_55utqx7tjrft5ojtbr67ypjdye	47	14	captures	capture	NNS
work_55utqx7tjrft5ojtbr67ypjdye	47	15	every	every	DT
work_55utqx7tjrft5ojtbr67ypjdye	47	16	English	English	NNP
work_55utqx7tjrft5ojtbr67ypjdye	47	17	morpho-	morpho-	NN
work_55utqx7tjrft5ojtbr67ypjdye	47	18	logical	logical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	47	19	exception	exception	NN
work_55utqx7tjrft5ojtbr67ypjdye	47	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	47	21	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	47	22	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	47	23	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	47	24	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	47	25	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	47	26	use	use	VB
work_55utqx7tjrft5ojtbr67ypjdye	47	27	context	context	NN
work_55utqx7tjrft5ojtbr67ypjdye	47	28	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	47	29	resolve	resolve	VB
work_55utqx7tjrft5ojtbr67ypjdye	47	30	ambiguous	ambiguous	JJ
work_55utqx7tjrft5ojtbr67ypjdye	47	31	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	47	32	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	47	33	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	48	1	They	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	48	2	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	48	3	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	48	4	deterministic	deterministic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	48	5	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	48	6	consistent	consistent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	48	7	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	48	8	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	48	9	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	48	10	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	48	11	if	if	IN
work_55utqx7tjrft5ojtbr67ypjdye	48	12	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	48	13	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	48	14	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	48	15	maps	map	NNS
work_55utqx7tjrft5ojtbr67ypjdye	48	16	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	48	17	stem	stem	VB
work_55utqx7tjrft5ojtbr67ypjdye	48	18	B	b	NN
work_55utqx7tjrft5ojtbr67ypjdye	48	19	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	48	20	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	48	21	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	48	22	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	48	23	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	48	24	will	will	MD
work_55utqx7tjrft5ojtbr67ypjdye	48	25	do	do	VB
work_55utqx7tjrft5ojtbr67ypjdye	48	26	so	so	RB
work_55utqx7tjrft5ojtbr67ypjdye	48	27	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	48	28	every	every	DT
work_55utqx7tjrft5ojtbr67ypjdye	48	29	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	48	30	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	48	31	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	48	32	A	A	NNP
work_55utqx7tjrft5ojtbr67ypjdye	48	33	arises	arise	NNS
work_55utqx7tjrft5ojtbr67ypjdye	48	34	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	49	1	Treat-	Treat-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	49	2	ments	ment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	49	4	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	49	5	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	6	therefore	therefore	RB
work_55utqx7tjrft5ojtbr67ypjdye	49	7	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	49	8	effectively	effectively	RB
work_55utqx7tjrft5ojtbr67ypjdye	49	9	equiva-	equiva-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	49	10	lence	lence	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	11	relations	relation	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	12	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	49	13	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	49	14	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	49	16	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	49	17	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	49	18	confla-	confla-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	49	19	tion	tion	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	20	class	class	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	21	being	be	VBG
work_55utqx7tjrft5ojtbr67ypjdye	49	22	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	49	23	equivalence	equivalence	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	24	class	class	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	25	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	49	26	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	27	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	28	under	under	IN
work_55utqx7tjrft5ojtbr67ypjdye	49	29	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	49	30	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	31	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	32	t.	t.	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	33	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	49	34	Jivani	Jivani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	49	35	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	49	36	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	49	37	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	49	38	refers	refer	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	49	39	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	49	40	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	49	41	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	49	42	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	49	43	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	44	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	45	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	49	46	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	49	47	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	49	48	affix	affix	NNP
work_55utqx7tjrft5ojtbr67ypjdye	49	49	removal	removal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	49	50	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	51	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	49	52	”	"	``
work_55utqx7tjrft5ojtbr67ypjdye	49	53	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	49	54	find	find	VBP
work_55utqx7tjrft5ojtbr67ypjdye	49	55	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	49	56	naming	name	VBG
work_55utqx7tjrft5ojtbr67ypjdye	49	57	confusing	confusing	NN
work_55utqx7tjrft5ojtbr67ypjdye	49	58	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	49	59	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	60	rarely	rarely	RB
work_55utqx7tjrft5ojtbr67ypjdye	49	61	strictly	strictly	RB
work_55utqx7tjrft5ojtbr67ypjdye	49	62	truncate	truncate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	49	63	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	49	64	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	49	65	almost	almost	RB
work_55utqx7tjrft5ojtbr67ypjdye	49	66	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	49	67	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	68	aim	aim	VBP
work_55utqx7tjrft5ojtbr67ypjdye	49	69	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	49	70	remove	remove	VB
work_55utqx7tjrft5ojtbr67ypjdye	49	71	affixes	affix	NNS
work_55utqx7tjrft5ojtbr67ypjdye	49	72	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	50	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	50	2	core	core	NN
work_55utqx7tjrft5ojtbr67ypjdye	50	3	similarity	similarity	NN
work_55utqx7tjrft5ojtbr67ypjdye	50	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	50	5	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	50	6	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	50	7	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	50	8	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	50	9	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	50	10	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	50	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	50	12	language	language	NN
work_55utqx7tjrft5ojtbr67ypjdye	50	13	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	50	14	specific	specific	JJ
work_55utqx7tjrft5ojtbr67ypjdye	50	15	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	50	16	used	use	VBN
work_55utqx7tjrft5ojtbr67ypjdye	50	17	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	50	18	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	50	19	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	50	20	mers	mer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	50	21	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	50	22	encoded	encode	VBN
work_55utqx7tjrft5ojtbr67ypjdye	50	23	directly	directly	RB
work_55utqx7tjrft5ojtbr67ypjdye	50	24	into	into	IN
work_55utqx7tjrft5ojtbr67ypjdye	50	25	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	50	26	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	50	27	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	50	28	apply	apply	VBP
work_55utqx7tjrft5ojtbr67ypjdye	50	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	51	1	Truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	51	2	Stemmers	Stemmers	NNP
work_55utqx7tjrft5ojtbr67ypjdye	51	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	52	1	k	k	LS
work_55utqx7tjrft5ojtbr67ypjdye	52	2	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	52	3	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	52	4	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	52	5	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	52	6	Bhamidipati	Bhamidipati	NNP
work_55utqx7tjrft5ojtbr67ypjdye	52	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	52	8	Pal	Pal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	52	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	52	10	2007	2007	CD
work_55utqx7tjrft5ojtbr67ypjdye	52	11	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	52	12	remove	remove	VB
work_55utqx7tjrft5ojtbr67ypjdye	52	13	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	52	14	but	but	IN
work_55utqx7tjrft5ojtbr67ypjdye	52	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	52	16	first	first	JJ
work_55utqx7tjrft5ojtbr67ypjdye	52	17	k	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	52	18	characters	character	NNS
work_55utqx7tjrft5ojtbr67ypjdye	52	19	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	52	20	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	52	21	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	52	22	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	53	1	As	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	53	2	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	53	3	naı̈ve	naı̈ve	NNP
work_55utqx7tjrft5ojtbr67ypjdye	53	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	53	5	high	high	JJ
work_55utqx7tjrft5ojtbr67ypjdye	53	6	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	53	7	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	53	8	method	method	NN
work_55utqx7tjrft5ojtbr67ypjdye	53	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	53	10	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	53	11	serve	serve	VBP
work_55utqx7tjrft5ojtbr67ypjdye	53	12	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	53	13	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	53	14	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	53	15	baseline	baseline	NN
work_55utqx7tjrft5ojtbr67ypjdye	53	16	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	53	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	53	18	rela-	rela-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	53	19	tive	tive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	53	20	effects	effect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	53	21	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	53	22	simple	simple	JJ
work_55utqx7tjrft5ojtbr67ypjdye	53	23	vocabulary	vocabulary	JJ
work_55utqx7tjrft5ojtbr67ypjdye	53	24	reduction	reduction	NN
work_55utqx7tjrft5ojtbr67ypjdye	53	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	54	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	54	2	test	test	VBP
work_55utqx7tjrft5ojtbr67ypjdye	54	3	four	four	CD
work_55utqx7tjrft5ojtbr67ypjdye	54	4	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	54	5	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	54	6	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	54	7	T4	t4	NN
work_55utqx7tjrft5ojtbr67ypjdye	54	8	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	54	9	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	54	10	five	five	CD
work_55utqx7tjrft5ojtbr67ypjdye	54	11	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	54	12	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	54	13	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	54	14	T5	T5	NNP
work_55utqx7tjrft5ojtbr67ypjdye	54	15	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	54	16	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	55	1	Five-	five-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	55	2	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	55	3	has	have	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	55	4	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	55	5	close	close	RB
work_55utqx7tjrft5ojtbr67ypjdye	55	6	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	55	7	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	55	8	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	55	9	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	55	10	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	55	11	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	55	12	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	55	13	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	55	14	levels	level	NNS
work_55utqx7tjrft5ojtbr67ypjdye	55	15	below	below	IN
work_55utqx7tjrft5ojtbr67ypjdye	55	16	four	four	CD
work_55utqx7tjrft5ojtbr67ypjdye	55	17	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	55	18	incoherent	incoherent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	55	19	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	56	1	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	56	2	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	56	3	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	56	4	Stemmer	Stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	56	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	57	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	57	2	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	57	3	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	57	4	removal	removal	NN
work_55utqx7tjrft5ojtbr67ypjdye	57	5	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	57	6	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	57	7	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	57	8	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	57	9	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	57	10	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	57	11	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	57	12	SS	SS	NNP
work_55utqx7tjrft5ojtbr67ypjdye	57	13	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	57	14	removes	remove	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	57	15	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	57	16	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	57	17	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	57	18	endings	ending	NNS
work_55utqx7tjrft5ojtbr67ypjdye	57	19	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	57	20	only	only	RB
work_55utqx7tjrft5ojtbr67ypjdye	57	21	three	three	CD
work_55utqx7tjrft5ojtbr67ypjdye	57	22	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	57	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	58	1	Harman	Harman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	58	2	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	58	3	1991	1991	CD
work_55utqx7tjrft5ojtbr67ypjdye	58	4	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	58	5	introduces	introduce	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	58	6	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	58	7	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	58	8	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	58	9	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	58	10	stem-	stem-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	58	11	ming	ming	NNP
work_55utqx7tjrft5ojtbr67ypjdye	58	12	algorithm	algorithm	NNP
work_55utqx7tjrft5ojtbr67ypjdye	58	13	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	58	14	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	58	15	weaker	weak	JJR
work_55utqx7tjrft5ojtbr67ypjdye	58	16	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	58	17	simpler	simple	JJR
work_55utqx7tjrft5ojtbr67ypjdye	58	18	counter-	counter-	XX
work_55utqx7tjrft5ojtbr67ypjdye	58	19	point	point	NN
work_55utqx7tjrft5ojtbr67ypjdye	58	20	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	58	21	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	58	22	standard	standard	JJ
work_55utqx7tjrft5ojtbr67ypjdye	58	23	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	58	24	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	58	25	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	58	26	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	58	27	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	59	1	As	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	59	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	59	3	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	59	4	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	59	5	simple	simple	JJ
work_55utqx7tjrft5ojtbr67ypjdye	59	6	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	59	7	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	59	8	representatives	representative	NNS
work_55utqx7tjrft5ojtbr67ypjdye	59	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	59	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	59	11	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	59	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	59	13	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	59	14	employed	employ	VBN
work_55utqx7tjrft5ojtbr67ypjdye	59	15	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	59	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	59	17	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	59	18	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	59	19	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	59	20	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	59	21	section	section	NN
work_55utqx7tjrft5ojtbr67ypjdye	59	22	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	59	23	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	59	24	include	include	VBP
work_55utqx7tjrft5ojtbr67ypjdye	59	25	them	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	59	26	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	59	27	Table	table	NN
work_55utqx7tjrft5ojtbr67ypjdye	59	28	1	1	CD
work_55utqx7tjrft5ojtbr67ypjdye	59	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	60	1	Lovins	Lovins	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	60	2	Stemmer	Stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	60	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	61	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	61	2	Lovins	Lovins	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	61	3	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	61	4	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	61	5	LS	LS	NNP
work_55utqx7tjrft5ojtbr67ypjdye	61	6	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	61	7	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	61	8	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	61	9	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	61	10	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	61	11	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	61	12	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	61	13	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	61	14	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	61	15	two	two	CD
work_55utqx7tjrft5ojtbr67ypjdye	61	16	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	61	17	step	step	NN
work_55utqx7tjrft5ojtbr67ypjdye	61	18	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	61	19	al-	al-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	61	20	gorithm	gorithm	NN
work_55utqx7tjrft5ojtbr67ypjdye	61	21	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	61	22	Lovins	lovin	NNS
work_55utqx7tjrft5ojtbr67ypjdye	61	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	61	24	1968	1968	CD
work_55utqx7tjrft5ojtbr67ypjdye	61	25	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	61	26	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	62	1	These	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	62	2	steps	step	NNS
work_55utqx7tjrft5ojtbr67ypjdye	62	3	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	62	4	long	long	JJ
work_55utqx7tjrft5ojtbr67ypjdye	62	5	lists	list	NNS
work_55utqx7tjrft5ojtbr67ypjdye	62	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	62	7	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	62	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	62	9	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	62	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	62	11	method	method	NN
work_55utqx7tjrft5ojtbr67ypjdye	62	12	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	62	13	still	still	RB
work_55utqx7tjrft5ojtbr67ypjdye	62	14	fast	fast	JJ
work_55utqx7tjrft5ojtbr67ypjdye	62	15	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	62	16	simple	simple	JJ
work_55utqx7tjrft5ojtbr67ypjdye	62	17	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	62	18	imple-	imple-	VB
work_55utqx7tjrft5ojtbr67ypjdye	62	19	288	288	CD
work_55utqx7tjrft5ojtbr67ypjdye	62	20	If	if	IN
work_55utqx7tjrft5ojtbr67ypjdye	62	21	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	62	22	ends	end	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	62	23	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	62	24	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	62	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	63	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	64	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	65	1	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	65	2	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	65	3	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	65	4	end	end	VB
work_55utqx7tjrft5ojtbr67ypjdye	65	5	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	65	6	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	65	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	66	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	67	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	68	1	replace	replace	VB
work_55utqx7tjrft5ojtbr67ypjdye	68	2	ending	end	VBG
work_55utqx7tjrft5ojtbr67ypjdye	68	3	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	68	4	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	68	5	-ies	-ies	``
work_55utqx7tjrft5ojtbr67ypjdye	68	6	-aies	-aie	NNS
work_55utqx7tjrft5ojtbr67ypjdye	68	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	68	8	-eies	-eies	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	68	9	-y	-y	FW
work_55utqx7tjrft5ojtbr67ypjdye	68	10	-es	-es	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	68	11	-aes	-aes	NN
work_55utqx7tjrft5ojtbr67ypjdye	68	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	68	13	-ees	-ees	JJR
work_55utqx7tjrft5ojtbr67ypjdye	68	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	68	15	-oes	-oes	:
work_55utqx7tjrft5ojtbr67ypjdye	68	16	-e	-e	:
work_55utqx7tjrft5ojtbr67ypjdye	68	17	-s	-s	:
work_55utqx7tjrft5ojtbr67ypjdye	68	18	-ss	-ss	''
work_55utqx7tjrft5ojtbr67ypjdye	68	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	68	20	-us	-us	:
work_55utqx7tjrft5ojtbr67ypjdye	68	21	-	-	:
work_55utqx7tjrft5ojtbr67ypjdye	68	22	Table	table	NN
work_55utqx7tjrft5ojtbr67ypjdye	68	23	1	1	CD
work_55utqx7tjrft5ojtbr67ypjdye	68	24	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	68	25	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	68	26	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	68	27	S	s	NN
work_55utqx7tjrft5ojtbr67ypjdye	68	28	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	68	29	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	68	30	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	68	31	Harman	Harman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	68	32	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	68	33	1991	1991	CD
work_55utqx7tjrft5ojtbr67ypjdye	68	34	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	68	35	consists	consist	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	68	36	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	68	37	three	three	CD
work_55utqx7tjrft5ojtbr67ypjdye	68	38	simple	simple	JJ
work_55utqx7tjrft5ojtbr67ypjdye	68	39	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	68	40	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	68	41	order	order	NN
work_55utqx7tjrft5ojtbr67ypjdye	68	42	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	69	1	Only	only	RB
work_55utqx7tjrft5ojtbr67ypjdye	69	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	69	3	first	first	JJ
work_55utqx7tjrft5ojtbr67ypjdye	69	4	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	69	5	applicable	applicable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	69	6	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	69	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	69	8	first	first	JJ
work_55utqx7tjrft5ojtbr67ypjdye	69	9	column	column	NN
work_55utqx7tjrft5ojtbr67ypjdye	69	10	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	69	11	applied	apply	VBN
work_55utqx7tjrft5ojtbr67ypjdye	69	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	70	1	ment	ment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	70	2	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	70	3	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	70	4	generally	generally	RB
work_55utqx7tjrft5ojtbr67ypjdye	70	5	considered	consider	VBN
work_55utqx7tjrft5ojtbr67ypjdye	70	6	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	70	7	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	70	8	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	70	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	71	1	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	71	2	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	71	3	Porter2	Porter2	NNP
work_55utqx7tjrft5ojtbr67ypjdye	71	4	Stemmers	Stemmers	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	71	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	72	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	72	2	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	72	3	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	72	4	mer	mer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	72	5	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	72	6	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	72	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	72	8	1980	1980	CD
work_55utqx7tjrft5ojtbr67ypjdye	72	9	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	72	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	72	11	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	72	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	72	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	72	14	most	most	RBS
work_55utqx7tjrft5ojtbr67ypjdye	72	15	popular	popular	JJ
work_55utqx7tjrft5ojtbr67ypjdye	72	16	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	72	17	cur-	cur-	DT
work_55utqx7tjrft5ojtbr67ypjdye	72	18	rent	rent	NN
work_55utqx7tjrft5ojtbr67ypjdye	72	19	use	use	NN
work_55utqx7tjrft5ojtbr67ypjdye	72	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	72	21	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	72	22	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	72	23	slightly	slightly	RB
work_55utqx7tjrft5ojtbr67ypjdye	72	24	less	less	RBR
work_55utqx7tjrft5ojtbr67ypjdye	72	25	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	72	26	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	72	27	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	72	28	intricate	intricate	JJ
work_55utqx7tjrft5ojtbr67ypjdye	72	29	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	72	30	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	72	31	Lovins	lovin	NNS
work_55utqx7tjrft5ojtbr67ypjdye	72	32	’	'	''
work_55utqx7tjrft5ojtbr67ypjdye	72	33	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	73	1	It	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	73	2	uses	use	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	73	3	five	five	CD
work_55utqx7tjrft5ojtbr67ypjdye	73	4	phases	phase	NNS
work_55utqx7tjrft5ojtbr67ypjdye	73	5	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	73	6	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	73	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	73	8	conditions	condition	NNS
work_55utqx7tjrft5ojtbr67ypjdye	73	9	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	73	10	match	match	VBP
work_55utqx7tjrft5ojtbr67ypjdye	73	11	patterns	pattern	NNS
work_55utqx7tjrft5ojtbr67ypjdye	73	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	73	13	vowel	vowel	NN
work_55utqx7tjrft5ojtbr67ypjdye	73	14	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	73	15	con-	con-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	73	16	sonant	sonant	NN
work_55utqx7tjrft5ojtbr67ypjdye	73	17	sequences	sequence	NNS
work_55utqx7tjrft5ojtbr67ypjdye	73	18	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	74	1	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	74	2	later	later	RB
work_55utqx7tjrft5ojtbr67ypjdye	74	3	created	create	VBD
work_55utqx7tjrft5ojtbr67ypjdye	74	4	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	74	5	slightly	slightly	RB
work_55utqx7tjrft5ojtbr67ypjdye	74	6	im-	im-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	74	7	proved	proved	JJ
work_55utqx7tjrft5ojtbr67ypjdye	74	8	version	version	NN
work_55utqx7tjrft5ojtbr67ypjdye	74	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	74	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	74	11	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	74	12	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	74	13	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	74	14	Snowball	Snowball	NNP
work_55utqx7tjrft5ojtbr67ypjdye	74	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	74	16	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	74	17	programming	programming	NN
work_55utqx7tjrft5ojtbr67ypjdye	74	18	language	language	NN
work_55utqx7tjrft5ojtbr67ypjdye	74	19	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	74	20	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	74	21	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	74	22	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	74	23	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	74	24	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	74	25	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	74	26	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	74	27	2001	2001	CD
work_55utqx7tjrft5ojtbr67ypjdye	74	28	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	74	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	75	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	75	2	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	75	3	both	both	CC
work_55utqx7tjrft5ojtbr67ypjdye	75	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	75	5	original	original	JJ
work_55utqx7tjrft5ojtbr67ypjdye	75	6	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	75	7	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	75	8	P1	p1	NN
work_55utqx7tjrft5ojtbr67ypjdye	75	9	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	75	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	75	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	75	12	new	new	JJ
work_55utqx7tjrft5ojtbr67ypjdye	75	13	version	version	NN
work_55utqx7tjrft5ojtbr67ypjdye	75	14	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	75	15	P2	P2	NNP
work_55utqx7tjrft5ojtbr67ypjdye	75	16	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	75	17	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	75	18	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	75	19	evaluation	evaluation	NN
work_55utqx7tjrft5ojtbr67ypjdye	75	20	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	76	1	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	76	2	/	/	SYM
work_55utqx7tjrft5ojtbr67ypjdye	76	3	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	76	4	Stemmer	Stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	76	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	77	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	77	2	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	77	3	/	/	SYM
work_55utqx7tjrft5ojtbr67ypjdye	77	4	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	77	5	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	77	6	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	77	7	PH	PH	NNP
work_55utqx7tjrft5ojtbr67ypjdye	77	8	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	77	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	77	10	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	77	11	Lancaster	Lancaster	NNP
work_55utqx7tjrft5ojtbr67ypjdye	77	12	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	77	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	77	14	iterates	iterate	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	77	15	indefinitely	indefinitely	RB
work_55utqx7tjrft5ojtbr67ypjdye	77	16	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	77	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	77	18	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	77	19	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	77	20	list	list	NN
work_55utqx7tjrft5ojtbr67ypjdye	77	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	77	22	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	77	23	some	some	DT
work_55utqx7tjrft5ojtbr67ypjdye	77	24	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	77	25	only	only	RB
work_55utqx7tjrft5ojtbr67ypjdye	77	26	ap-	ap-	XX
work_55utqx7tjrft5ojtbr67ypjdye	77	27	plying	ply	VBG
work_55utqx7tjrft5ojtbr67ypjdye	77	28	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	77	29	unmodified	unmodified	JJ
work_55utqx7tjrft5ojtbr67ypjdye	77	30	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	77	31	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	77	32	others	other	NNS
work_55utqx7tjrft5ojtbr67ypjdye	77	33	terminating	terminate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	77	34	iteration	iteration	NN
work_55utqx7tjrft5ojtbr67ypjdye	77	35	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	77	36	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	77	37	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	77	38	1990	1990	CD
work_55utqx7tjrft5ojtbr67ypjdye	77	39	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	77	40	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	78	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	78	2	slightly	slightly	RB
work_55utqx7tjrft5ojtbr67ypjdye	78	3	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	78	4	com-	com-	NN
work_55utqx7tjrft5ojtbr67ypjdye	78	5	plicated	plicate	VBD
work_55utqx7tjrft5ojtbr67ypjdye	78	6	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	78	7	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	78	8	structure	structure	NN
work_55utqx7tjrft5ojtbr67ypjdye	78	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	78	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	78	11	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	78	12	/	/	SYM
work_55utqx7tjrft5ojtbr67ypjdye	78	13	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	78	14	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	78	15	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	78	16	similar	similar	JJ
work_55utqx7tjrft5ojtbr67ypjdye	78	17	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	78	18	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	78	19	Lovins	Lovins	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	78	20	stemmer	stemmer	VB
work_55utqx7tjrft5ojtbr67ypjdye	78	21	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	78	22	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	78	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	79	1	2.2	2.2	CD
work_55utqx7tjrft5ojtbr67ypjdye	79	2	Context	Context	NNP
work_55utqx7tjrft5ojtbr67ypjdye	79	3	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	79	4	Based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	79	5	Treatments	Treatments	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	79	6	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	79	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	79	8	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	79	9	above	above	RB
work_55utqx7tjrft5ojtbr67ypjdye	79	10	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	79	11	fast	fast	JJ
work_55utqx7tjrft5ojtbr67ypjdye	79	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	79	13	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	79	14	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	79	15	impre-	impre-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	79	16	cise	cise	NNP
work_55utqx7tjrft5ojtbr67ypjdye	79	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	79	18	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	79	19	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	79	20	limited	limited	JJ
work_55utqx7tjrft5ojtbr67ypjdye	79	21	set	set	NN
work_55utqx7tjrft5ojtbr67ypjdye	79	22	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	79	23	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	79	24	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	79	25	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	79	26	account	account	VB
work_55utqx7tjrft5ojtbr67ypjdye	79	27	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	79	28	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	79	29	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	79	30	morphological	morphological	JJ
work_55utqx7tjrft5ojtbr67ypjdye	79	31	exceptions	exception	NNS
work_55utqx7tjrft5ojtbr67ypjdye	79	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	80	1	Subtleties	subtlety	NNS
work_55utqx7tjrft5ojtbr67ypjdye	80	2	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	80	3	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	80	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	80	5	difference	difference	NN
work_55utqx7tjrft5ojtbr67ypjdye	80	6	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	80	7	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	80	8	frosting	frost	VBG
work_55utqx7tjrft5ojtbr67ypjdye	80	9	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	80	10	windows	window	NNS
work_55utqx7tjrft5ojtbr67ypjdye	80	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	80	12	cake	cake	NN
work_55utqx7tjrft5ojtbr67ypjdye	80	13	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	80	14	frosting	frosting	NN
work_55utqx7tjrft5ojtbr67ypjdye	80	15	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	80	16	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	80	17	lost	lose	VBN
work_55utqx7tjrft5ojtbr67ypjdye	80	18	without	without	IN
work_55utqx7tjrft5ojtbr67ypjdye	80	19	contextual	contextual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	80	20	informa-	informa-	NN
work_55utqx7tjrft5ojtbr67ypjdye	80	21	tion	tion	NN
work_55utqx7tjrft5ojtbr67ypjdye	80	22	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	81	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	81	2	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	81	3	below	below	IN
work_55utqx7tjrft5ojtbr67ypjdye	81	4	use	use	NN
work_55utqx7tjrft5ojtbr67ypjdye	81	5	tools	tool	NNS
work_55utqx7tjrft5ojtbr67ypjdye	81	6	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	81	7	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	81	8	dictio-	dictio-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	81	9	naries	narie	NNS
work_55utqx7tjrft5ojtbr67ypjdye	81	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	81	11	inflectional	inflectional	JJ
work_55utqx7tjrft5ojtbr67ypjdye	81	12	analysis	analysis	NN
work_55utqx7tjrft5ojtbr67ypjdye	81	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	81	14	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	81	15	part	part	NN
work_55utqx7tjrft5ojtbr67ypjdye	81	16	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	81	17	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	81	18	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	81	19	speech	speech	NN
work_55utqx7tjrft5ojtbr67ypjdye	81	20	in-	in-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	81	21	ference	ference	NN
work_55utqx7tjrft5ojtbr67ypjdye	81	22	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	81	23	determine	determine	VB
work_55utqx7tjrft5ojtbr67ypjdye	81	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	81	25	correct	correct	JJ
work_55utqx7tjrft5ojtbr67ypjdye	81	26	conflated	conflate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	81	27	form	form	NN
work_55utqx7tjrft5ojtbr67ypjdye	81	28	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	81	29	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	81	30	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	81	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	82	1	As	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	82	2	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	82	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	82	4	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	82	5	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	82	6	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	82	7	consistently	consistently	RB
work_55utqx7tjrft5ojtbr67ypjdye	82	8	reduce	reduce	VB
work_55utqx7tjrft5ojtbr67ypjdye	82	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	82	10	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	82	11	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	82	12	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	82	13	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	82	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	82	15	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	82	16	form	form	NN
work_55utqx7tjrft5ojtbr67ypjdye	82	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	83	1	However	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	83	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	83	3	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	83	4	tools	tool	NNS
work_55utqx7tjrft5ojtbr67ypjdye	83	5	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	83	6	demand	demand	VBP
work_55utqx7tjrft5ojtbr67ypjdye	83	7	more	more	JJR
work_55utqx7tjrft5ojtbr67ypjdye	83	8	computational	computational	JJ
work_55utqx7tjrft5ojtbr67ypjdye	83	9	resources	resource	NNS
work_55utqx7tjrft5ojtbr67ypjdye	83	10	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	83	11	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	83	12	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	83	13	data	datum	NNS
work_55utqx7tjrft5ojtbr67ypjdye	83	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	83	15	lemmatizing	lemmatize	VBG
work_55utqx7tjrft5ojtbr67ypjdye	83	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	83	17	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	83	18	took	take	VBD
work_55utqx7tjrft5ojtbr67ypjdye	83	19	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	83	20	com-	com-	NN
work_55utqx7tjrft5ojtbr67ypjdye	83	21	putational	putational	JJ
work_55utqx7tjrft5ojtbr67ypjdye	83	22	time	time	NN
work_55utqx7tjrft5ojtbr67ypjdye	83	23	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	83	24	training	train	VBG
work_55utqx7tjrft5ojtbr67ypjdye	83	25	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	83	26	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	83	27	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	83	28	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	84	1	Krovetz	Krovetz	NNP
work_55utqx7tjrft5ojtbr67ypjdye	84	2	Stemmer	Stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	84	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	85	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	85	2	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	85	3	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	85	4	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	85	5	Krovetz	Krovetz	NNP
work_55utqx7tjrft5ojtbr67ypjdye	85	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	85	7	1993	1993	CD
work_55utqx7tjrft5ojtbr67ypjdye	85	8	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	85	9	uses	use	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	85	10	inflectional	inflectional	JJ
work_55utqx7tjrft5ojtbr67ypjdye	85	11	analysis	analysis	NN
work_55utqx7tjrft5ojtbr67ypjdye	85	12	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	85	13	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	85	14	dic-	dic-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	85	15	tionary	tionary	NN
work_55utqx7tjrft5ojtbr67ypjdye	85	16	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	85	17	determine	determine	VB
work_55utqx7tjrft5ojtbr67ypjdye	85	18	correct	correct	JJ
work_55utqx7tjrft5ojtbr67ypjdye	85	19	forms	form	NNS
work_55utqx7tjrft5ojtbr67ypjdye	85	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	85	21	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	85	22	before	before	IN
work_55utqx7tjrft5ojtbr67ypjdye	85	23	removing	remove	VBG
work_55utqx7tjrft5ojtbr67ypjdye	85	24	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	85	25	endings	ending	NNS
work_55utqx7tjrft5ojtbr67ypjdye	85	26	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	86	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	86	2	process	process	NN
work_55utqx7tjrft5ojtbr67ypjdye	86	3	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	86	4	complex	complex	JJ
work_55utqx7tjrft5ojtbr67ypjdye	86	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	86	6	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	86	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	86	8	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	86	9	itself	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	86	10	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	86	11	weak	weak	JJ
work_55utqx7tjrft5ojtbr67ypjdye	86	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	86	13	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	86	14	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	86	15	aims	aim	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	86	16	less	less	JJR
work_55utqx7tjrft5ojtbr67ypjdye	86	17	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	86	18	conflating	conflate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	86	19	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	86	20	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	86	21	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	86	22	parts	part	NNS
work_55utqx7tjrft5ojtbr67ypjdye	86	23	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	86	24	speech	speech	NN
work_55utqx7tjrft5ojtbr67ypjdye	86	25	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	86	26	normalizing	normalize	VBG
work_55utqx7tjrft5ojtbr67ypjdye	86	27	verb	verb	JJ
work_55utqx7tjrft5ojtbr67ypjdye	86	28	forms	form	NNS
work_55utqx7tjrft5ojtbr67ypjdye	86	29	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	86	30	removing	remove	VBG
work_55utqx7tjrft5ojtbr67ypjdye	86	31	pluralization	pluralization	NN
work_55utqx7tjrft5ojtbr67ypjdye	86	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	87	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	87	2	dictionary	dictionary	NN
work_55utqx7tjrft5ojtbr67ypjdye	87	3	itself	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	87	4	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	87	5	crucial	crucial	JJ
work_55utqx7tjrft5ojtbr67ypjdye	87	6	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	87	7	implementation	implementation	NN
work_55utqx7tjrft5ojtbr67ypjdye	87	8	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	87	9	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	87	10	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	87	11	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	87	12	stemmer	stemmer	JJ
work_55utqx7tjrft5ojtbr67ypjdye	87	13	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	87	14	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	87	15	KS	KS	NNP
work_55utqx7tjrft5ojtbr67ypjdye	87	16	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	87	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	87	18	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	87	19	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	87	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	87	21	Lemur	Lemur	NNP
work_55utqx7tjrft5ojtbr67ypjdye	87	22	Project	Project	NNP
work_55utqx7tjrft5ojtbr67ypjdye	87	23	implementation	implementation	NN
work_55utqx7tjrft5ojtbr67ypjdye	87	24	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	88	1	Lemmatizer	Lemmatizer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	88	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	89	1	Lemmatizers	lemmatizer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	89	2	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	89	3	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	89	4	database	database	NN
work_55utqx7tjrft5ojtbr67ypjdye	89	5	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	89	6	lem-	lem-	VBG
work_55utqx7tjrft5ojtbr67ypjdye	89	7	mas	mas	NNP
work_55utqx7tjrft5ojtbr67ypjdye	89	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	89	9	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	89	10	standardized	standardize	VBN
work_55utqx7tjrft5ojtbr67ypjdye	89	11	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	89	12	forms	form	NNS
work_55utqx7tjrft5ojtbr67ypjdye	89	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	89	14	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	89	15	order	order	NN
work_55utqx7tjrft5ojtbr67ypjdye	89	16	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	89	17	find	find	VB
work_55utqx7tjrft5ojtbr67ypjdye	89	18	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	89	19	best	good	JJS
work_55utqx7tjrft5ojtbr67ypjdye	89	20	normalized	normalized	JJ
work_55utqx7tjrft5ojtbr67ypjdye	89	21	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	89	22	form	form	NN
work_55utqx7tjrft5ojtbr67ypjdye	89	23	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	89	24	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	89	25	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	89	26	token	token	NN
work_55utqx7tjrft5ojtbr67ypjdye	89	27	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	90	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	90	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	90	3	method	method	NN
work_55utqx7tjrft5ojtbr67ypjdye	90	4	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	90	5	orders	order	NNS
work_55utqx7tjrft5ojtbr67ypjdye	90	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	90	7	magnitude	magnitude	NN
work_55utqx7tjrft5ojtbr67ypjdye	90	8	slower	slow	JJR
work_55utqx7tjrft5ojtbr67ypjdye	90	9	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	90	10	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	90	11	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	90	12	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	90	13	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	90	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	90	15	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	90	16	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	90	17	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	90	18	much	much	RB
work_55utqx7tjrft5ojtbr67ypjdye	90	19	more	more	JJR
work_55utqx7tjrft5ojtbr67ypjdye	90	20	princi-	princi-	NN
work_55utqx7tjrft5ojtbr67ypjdye	90	21	pled	plead	VBN
work_55utqx7tjrft5ojtbr67ypjdye	90	22	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	90	23	extremely	extremely	RB
work_55utqx7tjrft5ojtbr67ypjdye	90	24	unlikely	unlikely	JJ
work_55utqx7tjrft5ojtbr67ypjdye	90	25	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	90	26	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	90	27	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	90	28	conflate	conflate	NN
work_55utqx7tjrft5ojtbr67ypjdye	90	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	91	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	91	2	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	91	3	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	91	4	WordNet	WordNet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	91	6	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	91	7	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	91	8	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	91	9	WL	WL	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	10	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	91	11	implemented	implement	VBD
work_55utqx7tjrft5ojtbr67ypjdye	91	12	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	91	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	91	14	Natural	Natural	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	15	Language	Language	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	16	ToolKit	ToolKit	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	17	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	91	18	Bird	Bird	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	19	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	20	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	21	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	22	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	91	23	2009	2009	CD
work_55utqx7tjrft5ojtbr67ypjdye	91	24	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	91	25	along	along	IN
work_55utqx7tjrft5ojtbr67ypjdye	91	26	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	91	27	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	91	28	Stanford	Stanford	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	29	POS	POS	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	30	Tagger	Tagger	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	31	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	91	32	Toutanova	Toutanova	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	33	et	et	FW
work_55utqx7tjrft5ojtbr67ypjdye	91	34	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	35	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	91	36	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	91	37	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	91	38	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	91	39	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	91	40	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	91	41	unmodified	unmodified	JJ
work_55utqx7tjrft5ojtbr67ypjdye	91	42	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	91	43	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	91	44	provide	provide	VB
work_55utqx7tjrft5ojtbr67ypjdye	91	45	auxiliary	auxiliary	JJ
work_55utqx7tjrft5ojtbr67ypjdye	91	46	part	part	NN
work_55utqx7tjrft5ojtbr67ypjdye	91	47	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	91	48	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	91	49	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	91	50	speech	speech	NN
work_55utqx7tjrft5ojtbr67ypjdye	91	51	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	91	52	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	91	53	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	91	54	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	91	55	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	92	1	3	3	CD
work_55utqx7tjrft5ojtbr67ypjdye	92	2	Model	Model	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	3	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	92	4	Data	Data	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	92	5	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	92	6	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	92	7	paper	paper	NN
work_55utqx7tjrft5ojtbr67ypjdye	92	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	92	9	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	92	10	focus	focus	VBP
work_55utqx7tjrft5ojtbr67ypjdye	92	11	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	92	12	modeling	model	VBG
work_55utqx7tjrft5ojtbr67ypjdye	92	13	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	92	14	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	92	15	English	English	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	16	datasets	dataset	NNS
work_55utqx7tjrft5ojtbr67ypjdye	92	17	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	92	18	Latent	Latent	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	19	Dirichlet	Dirichlet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	20	Allocation	Allocation	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	21	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	92	22	LDA	LDA	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	23	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	92	24	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	92	25	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	92	26	generative	generative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	92	27	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	92	28	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	92	29	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	92	30	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	92	31	upon	upon	IN
work_55utqx7tjrft5ojtbr67ypjdye	92	32	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	92	33	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	92	34	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	92	35	Blei	Blei	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	36	et	et	FW
work_55utqx7tjrft5ojtbr67ypjdye	92	37	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	38	.	.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	92	39	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	92	40	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	92	41	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	92	42	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	93	1	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	93	2	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	93	3	φ	φ	NN
work_55utqx7tjrft5ojtbr67ypjdye	93	4	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	93	5	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	93	6	context	context	NN
work_55utqx7tjrft5ojtbr67ypjdye	93	7	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	93	8	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	93	9	multinomial	multinomial	JJ
work_55utqx7tjrft5ojtbr67ypjdye	93	10	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	93	11	distribution	distribution	NN
work_55utqx7tjrft5ojtbr67ypjdye	93	12	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	93	13	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	93	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	93	15	without	without	IN
work_55utqx7tjrft5ojtbr67ypjdye	93	16	any	any	DT
work_55utqx7tjrft5ojtbr67ypjdye	93	17	embedded	embed	VBN
work_55utqx7tjrft5ojtbr67ypjdye	93	18	semantic	semantic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	93	19	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	93	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	93	21	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	93	22	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	93	23	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	93	24	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	93	25	connected	connect	VBN
work_55utqx7tjrft5ojtbr67ypjdye	93	26	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	94	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	2	LDA	LDA	NNP
work_55utqx7tjrft5ojtbr67ypjdye	94	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	94	4	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	5	document	document	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	6	has	have	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	94	7	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	8	multinomial	multinomial	JJ
work_55utqx7tjrft5ojtbr67ypjdye	94	9	distribution	distribution	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	10	θ	θ	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	11	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	12	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	94	13	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	94	14	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	15	document	document	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	16	d	d	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	17	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	94	18	generated	generate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	94	19	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	20	choosing	choose	VBG
work_55utqx7tjrft5ojtbr67ypjdye	94	21	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	22	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	23	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	24	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	94	25	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	94	26	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	94	27	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	28	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	29	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	30	first	first	RB
work_55utqx7tjrft5ojtbr67ypjdye	94	31	sampling	sample	VBG
work_55utqx7tjrft5ojtbr67ypjdye	94	32	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	33	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	94	34	k	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	35	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	36	θd	θd	NNP
work_55utqx7tjrft5ojtbr67ypjdye	94	37	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	94	38	then	then	RB
work_55utqx7tjrft5ojtbr67ypjdye	94	39	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	40	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	41	w	w	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	42	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	43	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	44	distribution	distribution	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	45	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	46	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	94	47	φk	φk	CD
work_55utqx7tjrft5ojtbr67ypjdye	94	48	asso-	asso-	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	49	ciated	ciate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	94	50	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	51	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	52	k.	k.	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	53	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	54	name	name	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	55	Latent	Latent	NNP
work_55utqx7tjrft5ojtbr67ypjdye	94	56	Dirichlet	Dirichlet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	94	57	Allocation	Allocation	NNP
work_55utqx7tjrft5ojtbr67ypjdye	94	58	comes	come	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	94	59	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	60	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	61	assumptions	assumption	NNS
work_55utqx7tjrft5ojtbr67ypjdye	94	62	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	94	63	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	64	about	about	IN
work_55utqx7tjrft5ojtbr67ypjdye	94	65	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	66	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	67	’s	’s	POS
work_55utqx7tjrft5ojtbr67ypjdye	94	68	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	69	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	94	70	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	94	71	original	original	JJ
work_55utqx7tjrft5ojtbr67ypjdye	94	72	distributions	distribution	NNS
work_55utqx7tjrft5ojtbr67ypjdye	94	73	θ	θ	NNP
work_55utqx7tjrft5ojtbr67ypjdye	94	74	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	94	75	φ	φ	NNP
work_55utqx7tjrft5ojtbr67ypjdye	94	76	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	94	77	latent	latent	NN
work_55utqx7tjrft5ojtbr67ypjdye	94	78	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	94	79	i.e.	i.e.	FW
work_55utqx7tjrft5ojtbr67ypjdye	95	1	unobserved	unobserved	JJ
work_55utqx7tjrft5ojtbr67ypjdye	95	2	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	95	3	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	95	4	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	95	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	95	6	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	95	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	95	8	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	95	9	dis-	dis-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	10	tributions	tributions	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	11	θ	θ	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	12	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	95	13	φ	φ	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	14	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	95	15	drawn	draw	VBN
work_55utqx7tjrft5ojtbr67ypjdye	95	16	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	95	17	Dirichlet	Dirichlet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	18	distri-	distri-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	95	19	butions	bution	NNS
work_55utqx7tjrft5ojtbr67ypjdye	95	20	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	95	21	θ	θ	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	22	∼	∼	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	23	Dir(α	Dir(α	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	24	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	95	25	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	95	26	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	95	27	φ	φ	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	28	∼	∼	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	29	Dir(β	Dir(β	NNP
work_55utqx7tjrft5ojtbr67ypjdye	95	30	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	95	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	96	1	Using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	96	2	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	96	3	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	96	5	one	one	PRP
work_55utqx7tjrft5ojtbr67ypjdye	96	6	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	96	7	attempt	attempt	VB
work_55utqx7tjrft5ojtbr67ypjdye	96	8	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	96	9	infer	infer	VB
work_55utqx7tjrft5ojtbr67ypjdye	96	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	96	11	most	most	RBS
work_55utqx7tjrft5ojtbr67ypjdye	96	12	likely	likely	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	13	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	96	14	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	96	15	generate	generate	VB
work_55utqx7tjrft5ojtbr67ypjdye	96	16	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	96	17	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	18	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	96	19	some	some	DT
work_55utqx7tjrft5ojtbr67ypjdye	96	20	preset	preset	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	21	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	22	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	96	23	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	96	24	K.	K.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	96	25	However	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	96	26	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	96	27	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	96	28	optimization	optimization	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	29	problem	problem	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	30	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	96	31	non	non	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	32	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	33	convex	convex	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	34	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	96	35	intractable	intractable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	36	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	96	37	solve	solve	VB
work_55utqx7tjrft5ojtbr67ypjdye	96	38	analytically	analytically	RB
work_55utqx7tjrft5ojtbr67ypjdye	96	39	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	96	40	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	96	41	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	96	42	thus	thus	RB
work_55utqx7tjrft5ojtbr67ypjdye	96	43	generally	generally	RB
work_55utqx7tjrft5ojtbr67ypjdye	96	44	solved	solve	VBN
work_55utqx7tjrft5ojtbr67ypjdye	96	45	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	96	46	iterative	iterative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	47	techniques	technique	NNS
work_55utqx7tjrft5ojtbr67ypjdye	96	48	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	49	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	96	50	Gibbs	gibb	NNS
work_55utqx7tjrft5ojtbr67ypjdye	96	51	sampling	sampling	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	52	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	96	53	expectation	expectation	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	54	maximization	maximization	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	55	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	96	56	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	96	57	variational	variational	JJ
work_55utqx7tjrft5ojtbr67ypjdye	96	58	inference	inference	NN
work_55utqx7tjrft5ojtbr67ypjdye	96	59	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	97	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	97	2	resulting	result	VBG
work_55utqx7tjrft5ojtbr67ypjdye	97	3	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	97	4	fre-	fre-	NN
work_55utqx7tjrft5ojtbr67ypjdye	97	5	quently	quently	RB
work_55utqx7tjrft5ojtbr67ypjdye	97	6	display	display	VB
work_55utqx7tjrft5ojtbr67ypjdye	97	7	themes	theme	NNS
work_55utqx7tjrft5ojtbr67ypjdye	97	8	within	within	IN
work_55utqx7tjrft5ojtbr67ypjdye	97	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	97	10	common	common	JJ
work_55utqx7tjrft5ojtbr67ypjdye	97	11	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	97	12	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	97	13	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	97	14	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	97	15	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	97	16	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	97	17	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	97	18	used	use	VBN
work_55utqx7tjrft5ojtbr67ypjdye	97	19	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	97	20	classification	classification	NN
work_55utqx7tjrft5ojtbr67ypjdye	97	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	97	22	search	search	NN
work_55utqx7tjrft5ojtbr67ypjdye	97	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	97	24	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	97	25	recommendation	recommendation	NN
work_55utqx7tjrft5ojtbr67ypjdye	97	26	systems	system	NNS
work_55utqx7tjrft5ojtbr67ypjdye	97	27	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	98	1	Because	because	IN
work_55utqx7tjrft5ojtbr67ypjdye	98	2	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	98	3	affects	affect	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	98	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	98	5	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	98	6	distri-	distri-	NN
work_55utqx7tjrft5ojtbr67ypjdye	98	7	289	289	CD
work_55utqx7tjrft5ojtbr67ypjdye	98	8	bution	bution	NN
work_55utqx7tjrft5ojtbr67ypjdye	98	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	98	10	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	98	11	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	98	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	98	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	98	14	optimal	optimal	JJ
work_55utqx7tjrft5ojtbr67ypjdye	98	15	parameters	parameter	NNS
work_55utqx7tjrft5ojtbr67ypjdye	98	16	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	98	17	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	98	18	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	98	19	inference	inference	NN
work_55utqx7tjrft5ojtbr67ypjdye	98	20	will	will	MD
work_55utqx7tjrft5ojtbr67ypjdye	98	21	vary	vary	VB
work_55utqx7tjrft5ojtbr67ypjdye	98	22	depending	depend	VBG
work_55utqx7tjrft5ojtbr67ypjdye	98	23	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	98	24	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	98	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	99	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	99	2	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	99	3	adaptive	adaptive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	99	4	optimization	optimization	NN
work_55utqx7tjrft5ojtbr67ypjdye	99	5	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	99	6	both	both	CC
work_55utqx7tjrft5ojtbr67ypjdye	99	7	Dirichlet	Dirichlet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	99	8	hy-	hy-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	99	9	perparameters	perparameter	NNS
work_55utqx7tjrft5ojtbr67ypjdye	99	10	α	α	NNP
work_55utqx7tjrft5ojtbr67ypjdye	99	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	99	12	β	β	NNP
work_55utqx7tjrft5ojtbr67ypjdye	99	13	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	100	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	100	2	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	100	3	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	100	4	asymmetric	asymmetric	NN
work_55utqx7tjrft5ojtbr67ypjdye	100	5	α	α	NN
work_55utqx7tjrft5ojtbr67ypjdye	100	6	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	100	7	symmetric	symmetric	JJ
work_55utqx7tjrft5ojtbr67ypjdye	100	8	β	β	NNP
work_55utqx7tjrft5ojtbr67ypjdye	100	9	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	100	10	obtain	obtain	VB
work_55utqx7tjrft5ojtbr67ypjdye	100	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	100	12	best	good	JJS
work_55utqx7tjrft5ojtbr67ypjdye	100	13	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	100	14	fit	fit	NN
work_55utqx7tjrft5ojtbr67ypjdye	100	15	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	100	16	ac-	ac-	DT
work_55utqx7tjrft5ojtbr67ypjdye	100	17	cordance	cordance	NN
work_55utqx7tjrft5ojtbr67ypjdye	100	18	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	100	19	Wallach	Wallach	NNP
work_55utqx7tjrft5ojtbr67ypjdye	100	20	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	100	21	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	100	22	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	101	1	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	101	2	2009	2009	CD
work_55utqx7tjrft5ojtbr67ypjdye	101	3	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	101	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	102	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	102	2	order	order	NN
work_55utqx7tjrft5ojtbr67ypjdye	102	3	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	102	4	test	test	VB
work_55utqx7tjrft5ojtbr67ypjdye	102	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	102	6	various	various	JJ
work_55utqx7tjrft5ojtbr67ypjdye	102	7	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	102	8	normalization	normalization	NN
work_55utqx7tjrft5ojtbr67ypjdye	102	9	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	102	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	102	11	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	102	12	used	use	VBD
work_55utqx7tjrft5ojtbr67ypjdye	102	13	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	102	14	existing	exist	VBG
work_55utqx7tjrft5ojtbr67ypjdye	102	15	Python	Python	NNP
work_55utqx7tjrft5ojtbr67ypjdye	102	16	library	library	NN
work_55utqx7tjrft5ojtbr67ypjdye	102	17	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	102	18	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	102	19	Lovins	Lovins	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	102	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	102	21	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	102	22	/	/	SYM
work_55utqx7tjrft5ojtbr67ypjdye	102	23	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	102	24	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	102	25	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	102	26	both	both	CC
work_55utqx7tjrft5ojtbr67ypjdye	102	27	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	102	28	algorithms	algorithm	NNS
work_55utqx7tjrft5ojtbr67ypjdye	102	29	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	102	30	Chaput	Chaput	NNP
work_55utqx7tjrft5ojtbr67ypjdye	102	31	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	102	32	2010	2010	CD
work_55utqx7tjrft5ojtbr67ypjdye	102	33	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	102	34	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	102	35	modified	modify	VBN
work_55utqx7tjrft5ojtbr67ypjdye	102	36	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	102	37	correct	correct	VB
work_55utqx7tjrft5ojtbr67ypjdye	102	38	errors	error	NNS
work_55utqx7tjrft5ojtbr67ypjdye	102	39	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	102	40	im-	im-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	102	41	plementation	plementation	NN
work_55utqx7tjrft5ojtbr67ypjdye	102	42	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	103	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	103	2	implemented	implement	VBD
work_55utqx7tjrft5ojtbr67ypjdye	103	3	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	103	4	own	own	JJ
work_55utqx7tjrft5ojtbr67ypjdye	103	5	trunca-	trunca-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	103	6	tion	tion	NN
work_55utqx7tjrft5ojtbr67ypjdye	103	7	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	103	8	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	103	9	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	103	10	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	103	11	removal	removal	NN
work_55utqx7tjrft5ojtbr67ypjdye	103	12	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	103	13	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	104	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	104	2	applied	apply	VBD
work_55utqx7tjrft5ojtbr67ypjdye	104	3	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	104	4	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	5	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	6	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	104	7	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	8	token	token	VBN
work_55utqx7tjrft5ojtbr67ypjdye	104	9	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	10	four	four	CD
work_55utqx7tjrft5ojtbr67ypjdye	104	11	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	12	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	104	13	articles	article	NNS
work_55utqx7tjrft5ojtbr67ypjdye	104	14	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	15	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	16	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	17	early	early	JJ
work_55utqx7tjrft5ojtbr67ypjdye	104	18	2015,2	2015,2	CD
work_55utqx7tjrft5ojtbr67ypjdye	104	19	articles	article	NNS
work_55utqx7tjrft5ojtbr67ypjdye	104	20	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	21	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	104	22	New	New	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	23	York	York	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	24	Times	Times	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	25	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	26	2007	2007	CD
work_55utqx7tjrft5ojtbr67ypjdye	104	27	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	104	28	Sandhaus	Sandhaus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	29	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	104	30	2008	2008	CD
work_55utqx7tjrft5ojtbr67ypjdye	104	31	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	104	32	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	104	33	bi-	bi-	VB
work_55utqx7tjrft5ojtbr67ypjdye	104	34	ographies	ographie	NNS
work_55utqx7tjrft5ojtbr67ypjdye	104	35	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	36	IMDb,3	IMDb,3	NNS
work_55utqx7tjrft5ojtbr67ypjdye	104	37	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	104	38	reviews	review	NNS
work_55utqx7tjrft5ojtbr67ypjdye	104	39	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	40	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	104	41	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	42	Dataset	Dataset	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	43	Challenge.4	Challenge.4	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	44	Corpora	Corpora	NNP
work_55utqx7tjrft5ojtbr67ypjdye	104	45	were	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	104	46	partitioned	partition	VBN
work_55utqx7tjrft5ojtbr67ypjdye	104	47	into	into	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	48	75	75	CD
work_55utqx7tjrft5ojtbr67ypjdye	104	49	%	%	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	50	training	training	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	51	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	104	52	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	104	53	25	25	CD
work_55utqx7tjrft5ojtbr67ypjdye	104	54	%	%	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	55	test	test	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	56	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	104	57	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	104	58	lower	lower	RBR
work_55utqx7tjrft5ojtbr67ypjdye	104	59	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	104	60	cased	case	VBN
work_55utqx7tjrft5ojtbr67ypjdye	104	61	before	before	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	62	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	63	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	104	64	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	104	65	was	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	104	66	performed	perform	VBN
work_55utqx7tjrft5ojtbr67ypjdye	104	67	per	per	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	68	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	104	69	sentence	sentence	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	70	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	104	71	lower	lower	RBR
work_55utqx7tjrft5ojtbr67ypjdye	104	72	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	104	73	cased	case	VBN
work_55utqx7tjrft5ojtbr67ypjdye	104	74	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	104	75	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	105	1	After	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	105	2	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	105	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	105	4	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	105	5	remove	remove	VBP
work_55utqx7tjrft5ojtbr67ypjdye	105	6	stopwords	stopword	NNS
work_55utqx7tjrft5ojtbr67ypjdye	105	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	105	8	digits	digit	NNS
work_55utqx7tjrft5ojtbr67ypjdye	105	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	105	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	105	11	punctuation	punctuation	NN
work_55utqx7tjrft5ojtbr67ypjdye	105	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	106	1	Ta-	Ta-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	106	2	ble	ble	NNP
work_55utqx7tjrft5ojtbr67ypjdye	106	3	2	2	CD
work_55utqx7tjrft5ojtbr67ypjdye	106	4	shows	show	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	106	5	details	detail	NNS
work_55utqx7tjrft5ojtbr67ypjdye	106	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	106	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	106	8	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	106	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	106	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	106	11	Table	table	NN
work_55utqx7tjrft5ojtbr67ypjdye	106	12	3	3	CD
work_55utqx7tjrft5ojtbr67ypjdye	106	13	shows	show	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	106	14	examples	example	NNS
work_55utqx7tjrft5ojtbr67ypjdye	106	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	106	16	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	106	17	treatment.5	treatment.5	UH
work_55utqx7tjrft5ojtbr67ypjdye	106	18	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	106	19	train	train	VBP
work_55utqx7tjrft5ojtbr67ypjdye	106	20	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	106	21	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	106	22	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	106	23	MALLET	MALLET	NNP
work_55utqx7tjrft5ojtbr67ypjdye	106	24	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	106	25	McCallum	McCallum	NNP
work_55utqx7tjrft5ojtbr67ypjdye	106	26	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	106	27	2002	2002	CD
work_55utqx7tjrft5ojtbr67ypjdye	106	28	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	106	29	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	106	30	K	K	NNP
work_55utqx7tjrft5ojtbr67ypjdye	106	31	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	106	32	10	10	CD
work_55utqx7tjrft5ojtbr67ypjdye	106	33	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	106	34	50	50	CD
work_55utqx7tjrft5ojtbr67ypjdye	106	35	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	106	36	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	106	37	200	200	CD
work_55utqx7tjrft5ojtbr67ypjdye	106	38	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	106	39	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	106	40	at	at	RB
work_55utqx7tjrft5ojtbr67ypjdye	106	41	least	least	RBS
work_55utqx7tjrft5ojtbr67ypjdye	106	42	nine	nine	CD
work_55utqx7tjrft5ojtbr67ypjdye	106	43	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	106	44	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	106	45	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	106	46	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	106	47	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	106	48	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	106	49	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	106	50	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	106	51	K	K	NNP
work_55utqx7tjrft5ojtbr67ypjdye	106	52	combination	combination	NN
work_55utqx7tjrft5ojtbr67ypjdye	106	53	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	107	1	Training	Training	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	2	Data	Data	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	107	3	Evaluation	Evaluation	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	4	Data	Data	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	5	Corpus	Corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	6	#	#	,
work_55utqx7tjrft5ojtbr67ypjdye	107	7	docs	doc	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	107	8	#	#	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	9	toks	tok	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	107	10	#	#	$
work_55utqx7tjrft5ojtbr67ypjdye	107	11	docs	doc	NNS
work_55utqx7tjrft5ojtbr67ypjdye	107	12	#	#	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	13	toks	tok	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	107	14	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	15	articles	article	NNS
work_55utqx7tjrft5ojtbr67ypjdye	107	16	17.1	17.1	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	17	K	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	18	58.4	58.4	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	19	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	20	5.7	5.7	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	21	K	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	22	19.5	19.5	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	23	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	24	IMDb	IMDb	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	25	bios	bio	NNS
work_55utqx7tjrft5ojtbr67ypjdye	107	26	84.6	84.6	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	27	K	K	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	28	9.13	9.13	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	29	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	30	28.2	28.2	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	31	K	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	32	3.05	3.05	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	33	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	34	NYT	NYT	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	35	articles	article	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	107	36	29.4	29.4	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	37	K	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	38	8.81	8.81	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	39	M	m	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	40	9.79	9.79	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	41	K	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	42	2.98	2.98	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	43	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	44	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	45	reviews	review	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	107	46	844	844	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	47	K	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	48	43.1	43.1	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	49	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	50	281	281	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	51	K	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	52	14.4	14.4	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	53	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	54	Table	table	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	55	2	2	CD
work_55utqx7tjrft5ojtbr67ypjdye	107	56	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	107	57	Training	training	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	58	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	107	59	test	test	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	60	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	61	represent	represent	VBP
work_55utqx7tjrft5ojtbr67ypjdye	107	62	considerable	considerable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	107	63	variance	variance	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	64	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	107	65	content	content	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	66	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	107	67	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	68	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	107	69	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	107	70	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	107	71	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	107	72	length	length	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	73	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	107	74	doc-	doc-	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	75	ument	ument	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	76	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	107	77	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	107	78	proportion	proportion	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	79	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	107	80	training	training	NN
work_55utqx7tjrft5ojtbr67ypjdye	107	81	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	107	82	test	test	VB
work_55utqx7tjrft5ojtbr67ypjdye	107	83	data	datum	NNS
work_55utqx7tjrft5ojtbr67ypjdye	107	84	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	108	1	4	4	LS
work_55utqx7tjrft5ojtbr67ypjdye	108	2	Evaluations	evaluation	NNS
work_55utqx7tjrft5ojtbr67ypjdye	108	3	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	108	4	order	order	NN
work_55utqx7tjrft5ojtbr67ypjdye	108	5	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	108	6	evaluate	evaluate	VB
work_55utqx7tjrft5ojtbr67ypjdye	108	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	108	8	differences	difference	NNS
work_55utqx7tjrft5ojtbr67ypjdye	108	9	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	108	10	confla-	confla-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	108	11	tion	tion	NN
work_55utqx7tjrft5ojtbr67ypjdye	108	12	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	108	13	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	108	14	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	108	15	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	108	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	108	17	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	108	18	want	want	VBP
work_55utqx7tjrft5ojtbr67ypjdye	108	19	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	108	20	look	look	VB
work_55utqx7tjrft5ojtbr67ypjdye	108	21	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	108	22	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	108	23	variety	variety	NN
work_55utqx7tjrft5ojtbr67ypjdye	108	24	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	108	25	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	108	26	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	108	27	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	108	28	evaluation	evaluation	NN
work_55utqx7tjrft5ojtbr67ypjdye	108	29	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	108	30	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	108	31	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	108	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	109	1	Unfortunately	unfortunately	RB
work_55utqx7tjrft5ojtbr67ypjdye	109	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	109	3	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	109	4	described	describe	VBN
work_55utqx7tjrft5ojtbr67ypjdye	109	5	later	later	RB
work_55utqx7tjrft5ojtbr67ypjdye	109	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	109	7	standard	standard	JJ
work_55utqx7tjrft5ojtbr67ypjdye	109	8	2Retrieved	2retrieve	VBN
work_55utqx7tjrft5ojtbr67ypjdye	109	9	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	109	10	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	109	11	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	109	12	http://www.arxiv.org	http://www.arxiv.org	NNP
work_55utqx7tjrft5ojtbr67ypjdye	109	13	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	109	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	110	1	3Courtesy	3courtesy	NN
work_55utqx7tjrft5ojtbr67ypjdye	110	2	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	110	3	IMDb	IMDb	NNS
work_55utqx7tjrft5ojtbr67ypjdye	110	4	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	110	5	http://www.imdb.com	http://www.imdb.com	ADD
work_55utqx7tjrft5ojtbr67ypjdye	110	6	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	110	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	111	1	4Retrieved	4retrieved	JJ
work_55utqx7tjrft5ojtbr67ypjdye	111	2	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	111	3	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	111	4	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	111	5	http://www.yelp.com/	http://www.yelp.com/	NNP
work_55utqx7tjrft5ojtbr67ypjdye	111	6	dataset_challenge	dataset_challenge	NNP
work_55utqx7tjrft5ojtbr67ypjdye	111	7	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	111	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	112	1	5Our	5our	CD
work_55utqx7tjrft5ojtbr67ypjdye	112	2	code	code	NN
work_55utqx7tjrft5ojtbr67ypjdye	112	3	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	112	4	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	112	5	found	find	VBN
work_55utqx7tjrft5ojtbr67ypjdye	112	6	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	112	7	https://github.com/	https://github.com/	CD
work_55utqx7tjrft5ojtbr67ypjdye	112	8	heraldicsandfox	heraldicsandfox	NNP
work_55utqx7tjrft5ojtbr67ypjdye	112	9	/	/	SYM
work_55utqx7tjrft5ojtbr67ypjdye	112	10	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	112	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	113	1	evaluations	evaluation	NNS
work_55utqx7tjrft5ojtbr67ypjdye	113	2	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	113	3	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	113	4	quality	quality	NN
work_55utqx7tjrft5ojtbr67ypjdye	113	5	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	113	6	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	113	7	held	hold	VBN
work_55utqx7tjrft5ojtbr67ypjdye	113	8	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	113	9	out	out	RP
work_55utqx7tjrft5ojtbr67ypjdye	113	10	likeli-	likeli-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	113	11	hood	hood	NN
work_55utqx7tjrft5ojtbr67ypjdye	113	12	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	113	13	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	113	14	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	113	15	implicitly	implicitly	RB
work_55utqx7tjrft5ojtbr67ypjdye	113	16	affected	affect	VBN
work_55utqx7tjrft5ojtbr67ypjdye	113	17	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	113	18	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	113	19	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	113	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	113	21	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	113	22	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	113	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	114	1	To	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	114	2	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	114	3	able	able	JJ
work_55utqx7tjrft5ojtbr67ypjdye	114	4	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	114	5	compare	compare	VB
work_55utqx7tjrft5ojtbr67ypjdye	114	6	dif-	dif-	RP
work_55utqx7tjrft5ojtbr67ypjdye	114	7	ferent	ferent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	114	8	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	114	9	without	without	IN
work_55utqx7tjrft5ojtbr67ypjdye	114	10	simply	simply	RB
work_55utqx7tjrft5ojtbr67ypjdye	114	11	favoring	favor	VBG
work_55utqx7tjrft5ojtbr67ypjdye	114	12	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	114	13	maxi-	maxi-	NN
work_55utqx7tjrft5ojtbr67ypjdye	114	14	mum	mum	NNP
work_55utqx7tjrft5ojtbr67ypjdye	114	15	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	114	16	vocabulary	vocabulary	JJ
work_55utqx7tjrft5ojtbr67ypjdye	114	17	reduction	reduction	NN
work_55utqx7tjrft5ojtbr67ypjdye	114	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	114	19	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	114	20	create	create	VBP
work_55utqx7tjrft5ojtbr67ypjdye	114	21	mod-	mod-	NN
work_55utqx7tjrft5ojtbr67ypjdye	114	22	ified	ifie	VBD
work_55utqx7tjrft5ojtbr67ypjdye	114	23	versions	version	NNS
work_55utqx7tjrft5ojtbr67ypjdye	114	24	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	114	25	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	114	26	existing	exist	VBG
work_55utqx7tjrft5ojtbr67ypjdye	114	27	classic	classic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	114	28	evaluations	evaluation	NNS
work_55utqx7tjrft5ojtbr67ypjdye	114	29	as	as	RB
work_55utqx7tjrft5ojtbr67ypjdye	114	30	well	well	RB
work_55utqx7tjrft5ojtbr67ypjdye	114	31	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	114	32	new	new	JJ
work_55utqx7tjrft5ojtbr67ypjdye	114	33	metrics	metric	NNS
work_55utqx7tjrft5ojtbr67ypjdye	114	34	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	114	35	understanding	understand	VBG
work_55utqx7tjrft5ojtbr67ypjdye	114	36	differences	difference	NNS
work_55utqx7tjrft5ojtbr67ypjdye	114	37	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	114	38	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	114	39	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	114	40	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	114	41	level	level	NN
work_55utqx7tjrft5ojtbr67ypjdye	114	42	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	114	43	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	114	44	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	114	45	instead	instead	RB
work_55utqx7tjrft5ojtbr67ypjdye	114	46	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	114	47	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	114	48	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	115	1	4.1	4.1	CD
work_55utqx7tjrft5ojtbr67ypjdye	115	2	Held	hold	VBN
work_55utqx7tjrft5ojtbr67ypjdye	115	3	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	115	4	Out	out	RP
work_55utqx7tjrft5ojtbr67ypjdye	115	5	Likelihood	Likelihood	NNP
work_55utqx7tjrft5ojtbr67ypjdye	115	6	Strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	115	7	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	115	8	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	115	9	improve	improve	VB
work_55utqx7tjrft5ojtbr67ypjdye	115	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	115	11	joint	joint	JJ
work_55utqx7tjrft5ojtbr67ypjdye	115	12	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	115	13	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	115	14	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	115	15	occurring	occur	VBG
work_55utqx7tjrft5ojtbr67ypjdye	115	16	without	without	IN
work_55utqx7tjrft5ojtbr67ypjdye	115	17	improving	improve	VBG
work_55utqx7tjrft5ojtbr67ypjdye	115	18	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	115	19	qual-	qual-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	115	20	ity	ity	NN
work_55utqx7tjrft5ojtbr67ypjdye	115	21	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	115	22	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	115	23	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	115	24	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	116	1	As	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	116	2	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	116	3	reduce	reduce	VBP
work_55utqx7tjrft5ojtbr67ypjdye	116	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	116	5	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	116	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	116	8	vo-	vo-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	116	9	cabulary	cabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	116	11	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	116	12	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	116	13	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	116	14	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	15	distribution	distribution	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	16	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	116	17	spread	spread	VBN
work_55utqx7tjrft5ojtbr67ypjdye	116	18	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	116	19	fewer	few	JJR
work_55utqx7tjrft5ojtbr67ypjdye	116	20	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	116	21	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	116	22	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	116	23	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	116	24	its	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	116	25	extreme	extreme	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	26	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	116	27	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	116	28	probabil-	probabil-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	116	29	ity	ity	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	30	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	116	31	any	any	DT
work_55utqx7tjrft5ojtbr67ypjdye	116	32	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	33	under	under	IN
work_55utqx7tjrft5ojtbr67ypjdye	116	34	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	116	35	zero	zero	CD
work_55utqx7tjrft5ojtbr67ypjdye	116	36	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	116	37	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	38	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	116	39	would	would	MD
work_55utqx7tjrft5ojtbr67ypjdye	116	40	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	116	41	1.0	1.0	CD
work_55utqx7tjrft5ojtbr67ypjdye	116	42	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	117	1	Experiments	experiment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	117	2	confirmed	confirm	VBD
work_55utqx7tjrft5ojtbr67ypjdye	117	3	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	117	4	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	117	5	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	117	6	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	117	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	117	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	117	9	standard	standard	JJ
work_55utqx7tjrft5ojtbr67ypjdye	117	10	held	hold	VBN
work_55utqx7tjrft5ojtbr67ypjdye	117	11	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	117	12	out	out	RP
work_55utqx7tjrft5ojtbr67ypjdye	117	13	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	117	14	score	score	NN
work_55utqx7tjrft5ojtbr67ypjdye	117	15	L	L	NNP
work_55utqx7tjrft5ojtbr67ypjdye	117	16	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	117	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	117	18	test	test	NN
work_55utqx7tjrft5ojtbr67ypjdye	117	19	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	117	20	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	117	21	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	117	22	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	117	23	trained	train	VBN
work_55utqx7tjrft5ojtbr67ypjdye	117	24	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	117	25	ordered	order	VBD
work_55utqx7tjrft5ojtbr67ypjdye	117	26	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	117	27	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	117	28	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	117	29	much	much	JJ
work_55utqx7tjrft5ojtbr67ypjdye	117	30	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	117	31	reduce	reduce	VBP
work_55utqx7tjrft5ojtbr67ypjdye	117	32	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	117	33	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	117	34	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	117	35	assigning	assign	VBG
work_55utqx7tjrft5ojtbr67ypjdye	117	36	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	117	37	highest	high	JJS
work_55utqx7tjrft5ojtbr67ypjdye	117	38	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	117	39	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	117	40	those	those	DT
work_55utqx7tjrft5ojtbr67ypjdye	117	41	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	117	42	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	117	43	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	117	44	smallest	small	JJS
work_55utqx7tjrft5ojtbr67ypjdye	117	45	vocabularies	vocabulary	NNS
work_55utqx7tjrft5ojtbr67ypjdye	117	46	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	118	1	To	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	118	2	account	account	VB
work_55utqx7tjrft5ojtbr67ypjdye	118	3	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	118	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	118	5	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	6	improvement	improvement	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	7	caused	cause	VBN
work_55utqx7tjrft5ojtbr67ypjdye	118	8	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	118	9	reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	118	10	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	11	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	118	13	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	118	14	normalize	normalize	VBP
work_55utqx7tjrft5ojtbr67ypjdye	118	15	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	118	16	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	17	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	118	18	K	K	NNP
work_55utqx7tjrft5ojtbr67ypjdye	118	19	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	118	20	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	118	21	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	118	22	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	23	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	118	24	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	118	25	smoothed	smoothed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	118	26	un-	un-	CD
work_55utqx7tjrft5ojtbr67ypjdye	118	27	igram	igram	NNP
work_55utqx7tjrft5ojtbr67ypjdye	118	28	language	language	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	29	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	30	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	118	31	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	118	32	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	118	33	β	β	NNP
work_55utqx7tjrft5ojtbr67ypjdye	118	34	parameter	parameter	NN
work_55utqx7tjrft5ojtbr67ypjdye	118	35	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	119	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	119	2	calculate	calculate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	119	3	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	119	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	119	5	normalized	normalized	JJ
work_55utqx7tjrft5ojtbr67ypjdye	119	6	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	119	7	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	119	8	Lnorm	lnorm	IN
work_55utqx7tjrft5ojtbr67ypjdye	119	9	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	119	10	per	per	IN
work_55utqx7tjrft5ojtbr67ypjdye	119	11	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	119	12	token	token	VBN
work_55utqx7tjrft5ojtbr67ypjdye	119	13	metric	metric	JJ
work_55utqx7tjrft5ojtbr67ypjdye	119	14	PTLLnorm	ptllnorm	NN
work_55utqx7tjrft5ojtbr67ypjdye	119	15	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	119	16	put	put	VB
work_55utqx7tjrft5ojtbr67ypjdye	119	17	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	119	18	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	119	19	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	119	20	lengths	length	NNS
work_55utqx7tjrft5ojtbr67ypjdye	119	21	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	119	22	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	119	23	comparable	comparable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	119	24	scale	scale	NN
work_55utqx7tjrft5ojtbr67ypjdye	119	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	120	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	120	2	com-	com-	RB
work_55utqx7tjrft5ojtbr67ypjdye	120	3	pute	pute	VBP
work_55utqx7tjrft5ojtbr67ypjdye	120	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	120	5	unigram	unigram	JJ
work_55utqx7tjrft5ojtbr67ypjdye	120	6	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	7	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	8	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	120	9	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	120	10	smoothed	smoothed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	120	11	multinomial	multinomial	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	12	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	120	13	prior	prior	JJ
work_55utqx7tjrft5ojtbr67ypjdye	120	14	β	β	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	120	16	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	17	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	120	18	instances	instance	NNS
work_55utqx7tjrft5ojtbr67ypjdye	120	19	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	120	20	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	21	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	22	w	w	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	23	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	120	24	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	120	25	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	26	nw	nw	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	27	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	120	28	vocabulary	vocabulary	JJ
work_55utqx7tjrft5ojtbr67ypjdye	120	29	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	30	W	w	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	31	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	120	32	total	total	JJ
work_55utqx7tjrft5ojtbr67ypjdye	120	33	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	120	34	count	count	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	35	N	N	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	36	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	120	37	Lunigram	Lunigram	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	38	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	120	39	∏	∏	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	40	j	j	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	41	∏	∏	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	42	i	i	PRP
work_55utqx7tjrft5ojtbr67ypjdye	120	43	nwij	nwij	VBP
work_55utqx7tjrft5ojtbr67ypjdye	120	44	+	+	CC
work_55utqx7tjrft5ojtbr67ypjdye	120	45	β	β	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	46	N	n	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	47	+	+	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	48	Wβ	wβ	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	49	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	120	50	1	1	LS
work_55utqx7tjrft5ojtbr67ypjdye	120	51	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	120	52	Lnorm	lnorm	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	53	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	120	54	L	L	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	55	/	/	SYM
work_55utqx7tjrft5ojtbr67ypjdye	120	56	Lunigram	Lunigram	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	57	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	120	58	2	2	LS
work_55utqx7tjrft5ojtbr67ypjdye	120	59	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	120	60	PTLLnorm	ptllnorm	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	61	=	=	NFP
work_55utqx7tjrft5ojtbr67ypjdye	120	62	log(Lnorm	log(lnorm	UH
work_55utqx7tjrft5ojtbr67ypjdye	120	63	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	120	64	N	n	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	65	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	120	66	logL	logl	ADD
work_55utqx7tjrft5ojtbr67ypjdye	120	67	N	N	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	68	−	−	NNP
work_55utqx7tjrft5ojtbr67ypjdye	120	69	log(Lunigram	log(lunigram	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	70	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	120	71	N	n	NN
work_55utqx7tjrft5ojtbr67ypjdye	120	72	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	121	1	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	121	2	3	3	LS
work_55utqx7tjrft5ojtbr67ypjdye	121	3	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	121	4	Our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	121	5	resulting	result	VBG
work_55utqx7tjrft5ojtbr67ypjdye	121	6	metric	metric	JJ
work_55utqx7tjrft5ojtbr67ypjdye	121	7	measures	measure	NNS
work_55utqx7tjrft5ojtbr67ypjdye	121	8	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	121	9	much	much	RB
work_55utqx7tjrft5ojtbr67ypjdye	121	10	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	121	11	aver-	aver-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	121	12	age	age	NN
work_55utqx7tjrft5ojtbr67ypjdye	121	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	121	14	introduction	introduction	NN
work_55utqx7tjrft5ojtbr67ypjdye	121	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	121	16	multiple	multiple	JJ
work_55utqx7tjrft5ojtbr67ypjdye	121	17	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	121	18	improves	improve	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	121	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	121	20	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	121	21	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	121	22	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	121	23	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	121	24	occurring	occurring	NN
work_55utqx7tjrft5ojtbr67ypjdye	121	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	122	1	4.2	4.2	CD
work_55utqx7tjrft5ojtbr67ypjdye	122	2	Topic	Topic	NNP
work_55utqx7tjrft5ojtbr67ypjdye	122	3	Coherence	Coherence	NNP
work_55utqx7tjrft5ojtbr67ypjdye	122	4	Though	though	IN
work_55utqx7tjrft5ojtbr67ypjdye	122	5	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	122	6	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	122	7	describes	describe	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	122	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	122	9	statistical	statistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	122	10	like-	like-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	122	11	lihood	lihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	122	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	122	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	122	14	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	122	15	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	122	16	generating	generate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	122	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	122	18	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	122	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	122	20	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	122	21	290	290	CD
work_55utqx7tjrft5ojtbr67ypjdye	122	22	http://www.arxiv.org	http://www.arxiv.org	NNS
work_55utqx7tjrft5ojtbr67ypjdye	122	23	http://www.imdb.com	http://www.imdb.com	ADD
work_55utqx7tjrft5ojtbr67ypjdye	122	24	http://www.yelp.com/dataset_challenge	http://www.yelp.com/dataset_challenge	NN
work_55utqx7tjrft5ojtbr67ypjdye	122	25	http://www.yelp.com/dataset_challenge	http://www.yelp.com/dataset_challenge	VBP
work_55utqx7tjrft5ojtbr67ypjdye	122	26	https://github.com/heraldicsandfox/stemmers	https://github.com/heraldicsandfox/stemmers	NNP
work_55utqx7tjrft5ojtbr67ypjdye	122	27	https://github.com/heraldicsandfox/stemmers	https://github.com/heraldicsandfox/stemmers	NNP
work_55utqx7tjrft5ojtbr67ypjdye	122	28	Original	Original	NNP
work_55utqx7tjrft5ojtbr67ypjdye	122	29	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	122	30	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	122	31	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	122	32	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	122	33	have	have	VB
work_55utqx7tjrft5ojtbr67ypjdye	122	34	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	122	35	service	service	NN
work_55utqx7tjrft5ojtbr67ypjdye	122	36	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	123	1	Went	go	VBD
work_55utqx7tjrft5ojtbr67ypjdye	123	2	through	through	IN
work_55utqx7tjrft5ojtbr67ypjdye	123	3	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	123	4	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	123	5	through	through	NN
work_55utqx7tjrft5ojtbr67ypjdye	123	6	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	123	7	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	123	8	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	123	9	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	123	10	drinks	drink	NNS
work_55utqx7tjrft5ojtbr67ypjdye	123	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	123	12	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	123	13	sides	side	NNS
work_55utqx7tjrft5ojtbr67ypjdye	123	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	124	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	124	2	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	124	3	were	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	124	4	preparing	prepare	VBG
work_55utqx7tjrft5ojtbr67ypjdye	124	5	what	what	WP
work_55utqx7tjrft5ojtbr67ypjdye	124	6	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	124	7	forgot	forgot	VBP
work_55utqx7tjrft5ojtbr67ypjdye	124	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	124	9	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	124	10	could	could	MD
work_55utqx7tjrft5ojtbr67ypjdye	124	11	see	see	VB
work_55utqx7tjrft5ojtbr67ypjdye	124	12	another	another	DT
work_55utqx7tjrft5ojtbr67ypjdye	124	13	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	124	14	who	who	WP
work_55utqx7tjrft5ojtbr67ypjdye	124	15	had	have	VBD
work_55utqx7tjrft5ojtbr67ypjdye	124	16	her	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	124	17	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	124	18	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	124	19	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	124	20	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	124	21	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	124	22	was	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	124	23	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	124	24	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	124	25	she	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	124	26	was	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	124	27	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	124	28	her	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	124	29	phone	phone	NN
work_55utqx7tjrft5ojtbr67ypjdye	124	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	125	1	Any	any	DT
work_55utqx7tjrft5ojtbr67ypjdye	125	2	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	125	3	KFC	kfc	NN
work_55utqx7tjrft5ojtbr67ypjdye	125	4	would	would	MD
work_55utqx7tjrft5ojtbr67ypjdye	125	5	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	125	6	better	well	JJR
work_55utqx7tjrft5ojtbr67ypjdye	125	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	126	1	Tokenized	tokenize	VBN
work_55utqx7tjrft5ojtbr67ypjdye	126	2	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	126	3	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	4	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	126	5	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	126	6	have	have	VB
work_55utqx7tjrft5ojtbr67ypjdye	126	7	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	8	service	service	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	9	went	go	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	10	through	through	IN
work_55utqx7tjrft5ojtbr67ypjdye	126	11	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	12	through	through	RB
work_55utqx7tjrft5ojtbr67ypjdye	126	13	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	126	14	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	126	15	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	16	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	126	17	drinks	drink	NNS
work_55utqx7tjrft5ojtbr67ypjdye	126	18	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	126	19	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	126	20	sides	side	NNS
work_55utqx7tjrft5ojtbr67ypjdye	126	21	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	126	22	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	126	23	were	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	24	preparing	prepare	VBG
work_55utqx7tjrft5ojtbr67ypjdye	126	25	what	what	WP
work_55utqx7tjrft5ojtbr67ypjdye	126	26	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	126	27	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	28	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	126	29	could	could	MD
work_55utqx7tjrft5ojtbr67ypjdye	126	30	see	see	VB
work_55utqx7tjrft5ojtbr67ypjdye	126	31	another	another	DT
work_55utqx7tjrft5ojtbr67ypjdye	126	32	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	33	who	who	WP
work_55utqx7tjrft5ojtbr67ypjdye	126	34	had	have	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	35	her	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	126	36	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	126	37	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	126	38	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	126	39	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	126	40	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	126	41	was	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	42	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	43	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	126	44	she	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	126	45	was	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	46	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	126	47	her	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	126	48	phone	phone	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	49	any	any	DT
work_55utqx7tjrft5ojtbr67ypjdye	126	50	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	51	kfc	kfc	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	52	would	would	MD
work_55utqx7tjrft5ojtbr67ypjdye	126	53	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	126	54	better	well	JJR
work_55utqx7tjrft5ojtbr67ypjdye	126	55	Stopped	stop	VBN
work_55utqx7tjrft5ojtbr67ypjdye	126	56	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	57	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	58	service	service	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	59	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	60	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	61	drinks	drink	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	126	62	sides	side	NNS
work_55utqx7tjrft5ojtbr67ypjdye	126	63	preparing	prepare	VBG
work_55utqx7tjrft5ojtbr67ypjdye	126	64	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	65	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	66	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	126	67	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	68	phone	phone	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	69	kfc	kfc	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	70	NS	NS	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	71	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	72	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	73	service	service	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	74	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	75	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	76	drinks	drink	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	126	77	sides	side	NNS
work_55utqx7tjrft5ojtbr67ypjdye	126	78	preparing	prepare	VBG
work_55utqx7tjrft5ojtbr67ypjdye	126	79	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	126	80	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	81	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	126	82	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	83	...	...	:
work_55utqx7tjrft5ojtbr67ypjdye	126	84	T4	t4	VB
work_55utqx7tjrft5ojtbr67ypjdye	126	85	loca	loca	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	86	good	good	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	87	serv	serv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	88	driv	driv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	89	forg	forg	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	90	drin	drin	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	91	side	side	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	92	prep	prep	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	93	forg	forg	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	94	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	95	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	126	96	obvi	obvi	RB
work_55utqx7tjrft5ojtbr67ypjdye	126	97	...	...	.
work_55utqx7tjrft5ojtbr67ypjdye	126	98	T5	t5	DT
work_55utqx7tjrft5ojtbr67ypjdye	126	99	locat	locat	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	100	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	126	101	servi	servi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	102	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	103	forgo	forgo	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	104	drink	drink	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	105	sides	side	NNS
work_55utqx7tjrft5ojtbr67ypjdye	126	106	prepa	prepa	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	107	forgo	forgo	NNP
work_55utqx7tjrft5ojtbr67ypjdye	126	108	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	126	109	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	126	110	obvio	obvio	NNS
work_55utqx7tjrft5ojtbr67ypjdye	126	111	...	...	.
work_55utqx7tjrft5ojtbr67ypjdye	127	1	LO	LO	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	2	loc	loc	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	3	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	4	servic	servic	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	5	dr	dr	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	6	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	7	drink	drink	VBP
work_55utqx7tjrft5ojtbr67ypjdye	127	8	sid	sid	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	9	prepar	prepar	NNS
work_55utqx7tjrft5ojtbr67ypjdye	127	10	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	11	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	12	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	127	13	obv	obv	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	14	...	...	:
work_55utqx7tjrft5ojtbr67ypjdye	127	15	P1	p1	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	16	locat	locat	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	17	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	18	servic	servic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	19	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	20	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	21	drink	drink	VBP
work_55utqx7tjrft5ojtbr67ypjdye	127	22	side	side	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	23	prepar	prepar	NNS
work_55utqx7tjrft5ojtbr67ypjdye	127	24	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	25	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	26	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	127	27	obviou	obviou	RB
work_55utqx7tjrft5ojtbr67ypjdye	127	28	...	...	NFP
work_55utqx7tjrft5ojtbr67ypjdye	127	29	P2	p2	VB
work_55utqx7tjrft5ojtbr67ypjdye	127	30	locat	locat	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	31	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	32	servic	servic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	33	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	34	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	35	drink	drink	VBP
work_55utqx7tjrft5ojtbr67ypjdye	127	36	side	side	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	37	prepar	prepar	NNS
work_55utqx7tjrft5ojtbr67ypjdye	127	38	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	39	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	40	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	127	41	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	42	...	...	:
work_55utqx7tjrft5ojtbr67ypjdye	127	43	PH	PH	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	44	loc	loc	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	45	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	46	serv	serv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	47	driv	driv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	48	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	49	drink	drink	VBP
work_55utqx7tjrft5ojtbr67ypjdye	127	50	sid	sid	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	51	prep	prep	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	52	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	53	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	54	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	127	55	obvy	obvy	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	56	...	...	:
work_55utqx7tjrft5ojtbr67ypjdye	127	57	SS	ss	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	58	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	59	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	60	service	service	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	61	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	62	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	63	drink	drink	VBP
work_55utqx7tjrft5ojtbr67ypjdye	127	64	side	side	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	65	preparing	prepare	VBG
work_55utqx7tjrft5ojtbr67ypjdye	127	66	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	67	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	68	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	127	69	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	70	...	...	:
work_55utqx7tjrft5ojtbr67ypjdye	127	71	KR	kr	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	72	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	73	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	74	service	service	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	75	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	76	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	77	drink	drink	VBP
work_55utqx7tjrft5ojtbr67ypjdye	127	78	side	side	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	79	prepare	prepare	VB
work_55utqx7tjrft5ojtbr67ypjdye	127	80	forgot	forget	VBD
work_55utqx7tjrft5ojtbr67ypjdye	127	81	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	82	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	127	83	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	84	...	...	:
work_55utqx7tjrft5ojtbr67ypjdye	127	85	WL	WL	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	86	location	location	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	87	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	88	service	service	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	89	drive	drive	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	90	forget	forget	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	91	drink	drink	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	92	side	side	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	93	prepare	prepare	VBP
work_55utqx7tjrft5ojtbr67ypjdye	127	94	forget	forget	VB
work_55utqx7tjrft5ojtbr67ypjdye	127	95	girl	girl	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	96	back	back	RB
work_55utqx7tjrft5ojtbr67ypjdye	127	97	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	127	98	...	...	:
work_55utqx7tjrft5ojtbr67ypjdye	127	99	Table	table	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	100	3	3	CD
work_55utqx7tjrft5ojtbr67ypjdye	127	101	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	127	102	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	127	103	demonstration	demonstration	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	104	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	127	105	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	127	106	steps	step	NNS
work_55utqx7tjrft5ojtbr67ypjdye	127	107	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	127	108	preprocessing	preprocesse	VBG
work_55utqx7tjrft5ojtbr67ypjdye	127	109	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	127	110	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	127	111	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	127	112	review	review	NN
work_55utqx7tjrft5ojtbr67ypjdye	127	113	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	128	1	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	128	2	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	128	3	necessarily	necessarily	RB
work_55utqx7tjrft5ojtbr67ypjdye	128	4	indicate	indicate	VB
work_55utqx7tjrft5ojtbr67ypjdye	128	5	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	128	6	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	128	7	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	128	8	seman-	seman-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	128	9	tically	tically	RB
work_55utqx7tjrft5ojtbr67ypjdye	128	10	coherent	coherent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	128	11	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	128	12	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	128	13	human	human	JJ
work_55utqx7tjrft5ojtbr67ypjdye	128	14	observer	observer	NN
work_55utqx7tjrft5ojtbr67ypjdye	128	15	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	129	1	To	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	129	2	measure	measure	VB
work_55utqx7tjrft5ojtbr67ypjdye	129	3	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	129	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	129	5	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	129	6	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	129	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	129	8	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	129	9	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	129	10	measure	measure	NN
work_55utqx7tjrft5ojtbr67ypjdye	129	11	proposed	propose	VBN
work_55utqx7tjrft5ojtbr67ypjdye	129	12	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	129	13	Mimno	Mimno	NNP
work_55utqx7tjrft5ojtbr67ypjdye	129	14	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	129	15	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	129	16	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	130	1	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	130	2	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	130	3	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	130	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	131	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	131	2	metric	metric	NN
work_55utqx7tjrft5ojtbr67ypjdye	131	3	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	131	4	defined	define	VBN
work_55utqx7tjrft5ojtbr67ypjdye	131	5	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	131	6	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	131	7	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	131	8	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	131	9	k	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	131	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	131	11	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	131	12	list	list	NN
work_55utqx7tjrft5ojtbr67ypjdye	131	13	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	131	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	131	15	top	top	JJ
work_55utqx7tjrft5ojtbr67ypjdye	131	16	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	131	17	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	131	18	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	131	19	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	131	20	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	131	21	vk1	vk1	NN
work_55utqx7tjrft5ojtbr67ypjdye	131	22	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	131	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	132	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	133	1	.v	.v	NFP
work_55utqx7tjrft5ojtbr67ypjdye	133	2	k	k	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	3	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	4	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	133	5	C(k	c(k	NN
work_55utqx7tjrft5ojtbr67ypjdye	133	6	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	133	7	=	=	NFP
work_55utqx7tjrft5ojtbr67ypjdye	133	8	M∑	M∑	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	9	m=2	m=2	NN
work_55utqx7tjrft5ojtbr67ypjdye	133	10	m−1∑	m−1∑	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	11	l=1	l=1	JJ
work_55utqx7tjrft5ojtbr67ypjdye	133	12	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	133	13	D(vl	D(vl	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	133	15	vm	vm	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	16	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	133	17	+	+	CC
work_55utqx7tjrft5ojtbr67ypjdye	133	18	β	β	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	19	D(vl	D(vl	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	20	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	133	21	+	+	SYM
work_55utqx7tjrft5ojtbr67ypjdye	133	22	β	β	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	23	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	133	24	4	4	CD
work_55utqx7tjrft5ojtbr67ypjdye	133	25	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	133	26	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	133	27	D(vl	D(vl	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	28	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	133	29	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	133	30	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	133	31	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	133	32	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	133	33	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	133	34	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	133	35	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	133	36	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	133	37	vl	vl	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	38	occurs	occur	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	133	39	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	133	40	D(vl	D(vl	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	41	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	133	42	vm	vm	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	43	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	133	44	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	133	45	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	133	46	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	133	47	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	133	48	doc-	doc-	NN
work_55utqx7tjrft5ojtbr67ypjdye	133	49	uments	ument	NNS
work_55utqx7tjrft5ojtbr67ypjdye	133	50	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	133	51	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	133	52	both	both	DT
work_55utqx7tjrft5ojtbr67ypjdye	133	53	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	133	54	vl	vl	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	55	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	133	56	vm	vm	NNP
work_55utqx7tjrft5ojtbr67ypjdye	133	57	occur	occur	VBP
work_55utqx7tjrft5ojtbr67ypjdye	133	58	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	134	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	134	2	metric	metric	NN
work_55utqx7tjrft5ojtbr67ypjdye	134	3	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	134	4	similar	similar	JJ
work_55utqx7tjrft5ojtbr67ypjdye	134	5	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	134	6	pointwise	pointwise	VB
work_55utqx7tjrft5ojtbr67ypjdye	134	7	mutual	mutual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	134	8	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	134	9	Lau	Lau	NNP
work_55utqx7tjrft5ojtbr67ypjdye	134	10	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	134	11	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	134	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	135	1	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	135	2	2014	2014	CD
work_55utqx7tjrft5ojtbr67ypjdye	135	3	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	135	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	135	5	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	135	6	instead	instead	RB
work_55utqx7tjrft5ojtbr67ypjdye	135	7	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	135	8	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	135	9	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	135	10	sliding	slide	VBG
work_55utqx7tjrft5ojtbr67ypjdye	135	11	win-	win-	NN
work_55utqx7tjrft5ojtbr67ypjdye	135	12	dow	dow	NNP
work_55utqx7tjrft5ojtbr67ypjdye	135	13	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	135	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	135	15	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	135	16	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	135	17	determine	determine	VB
work_55utqx7tjrft5ojtbr67ypjdye	135	18	co	co	NN
work_55utqx7tjrft5ojtbr67ypjdye	135	19	-	-	NN
work_55utqx7tjrft5ojtbr67ypjdye	135	20	occurrence	occurrence	NN
work_55utqx7tjrft5ojtbr67ypjdye	135	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	135	22	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	135	23	uses	use	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	135	24	full	full	JJ
work_55utqx7tjrft5ojtbr67ypjdye	135	25	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	135	26	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	135	27	discrete	discrete	JJ
work_55utqx7tjrft5ojtbr67ypjdye	135	28	windows	window	NNS
work_55utqx7tjrft5ojtbr67ypjdye	135	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	136	1	To	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	136	2	avoid	avoid	VB
work_55utqx7tjrft5ojtbr67ypjdye	136	3	biasing	bias	VBG
work_55utqx7tjrft5ojtbr67ypjdye	136	4	towards	towards	IN
work_55utqx7tjrft5ojtbr67ypjdye	136	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	136	6	smaller	small	JJR
work_55utqx7tjrft5ojtbr67ypjdye	136	7	vocabular-	vocabular-	NN
work_55utqx7tjrft5ojtbr67ypjdye	136	8	ies	ie	NNS
work_55utqx7tjrft5ojtbr67ypjdye	136	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	136	10	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	136	11	datasets	dataset	NNS
work_55utqx7tjrft5ojtbr67ypjdye	136	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	136	13	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	136	14	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	136	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	136	16	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	136	17	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	136	18	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	136	19	as-	as-	NN
work_55utqx7tjrft5ojtbr67ypjdye	136	20	signments	signment	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	136	21	output	output	NN
work_55utqx7tjrft5ojtbr67ypjdye	136	22	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	136	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	136	24	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	136	25	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	136	26	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	136	27	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	136	28	list	list	NN
work_55utqx7tjrft5ojtbr67ypjdye	136	29	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	136	30	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	136	31	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	136	32	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	136	33	produce	produce	VB
work_55utqx7tjrft5ojtbr67ypjdye	136	34	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	136	35	top	top	JJ
work_55utqx7tjrft5ojtbr67ypjdye	136	36	keywords	keyword	NNS
work_55utqx7tjrft5ojtbr67ypjdye	136	37	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	136	38	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	136	39	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	136	40	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	137	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	137	2	then	then	RB
work_55utqx7tjrft5ojtbr67ypjdye	137	3	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	137	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	137	5	original	original	JJ
work_55utqx7tjrft5ojtbr67ypjdye	137	6	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	137	7	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	137	8	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	137	9	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	137	10	new	new	JJ
work_55utqx7tjrft5ojtbr67ypjdye	137	11	keywords	keyword	NNS
work_55utqx7tjrft5ojtbr67ypjdye	137	12	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	137	13	compute	compute	VB
work_55utqx7tjrft5ojtbr67ypjdye	137	14	coher-	coher-	NN
work_55utqx7tjrft5ojtbr67ypjdye	137	15	ence	ence	NN
work_55utqx7tjrft5ojtbr67ypjdye	137	16	values	value	NNS
work_55utqx7tjrft5ojtbr67ypjdye	137	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	138	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	138	2	allows	allow	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	138	3	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	138	4	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	138	5	observe	observe	VB
work_55utqx7tjrft5ojtbr67ypjdye	138	6	whether	whether	IN
work_55utqx7tjrft5ojtbr67ypjdye	138	7	con-	con-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	138	8	flation	flation	NN
work_55utqx7tjrft5ojtbr67ypjdye	138	9	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	138	10	map	map	VBP
work_55utqx7tjrft5ojtbr67ypjdye	138	11	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	138	12	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	138	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	138	14	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	138	15	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	138	16	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	138	17	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	138	18	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	138	19	coherent	coherent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	138	20	way	way	NN
work_55utqx7tjrft5ojtbr67ypjdye	138	21	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	138	22	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	138	23	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	138	24	would	would	MD
work_55utqx7tjrft5ojtbr67ypjdye	138	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	139	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	139	2	experimented	experiment	VBD
work_55utqx7tjrft5ojtbr67ypjdye	139	3	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	139	4	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	139	5	Wikipedia	Wikipedia	NNP
work_55utqx7tjrft5ojtbr67ypjdye	139	6	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	139	7	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	139	8	refer-	refer-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	139	9	ence	ence	NN
work_55utqx7tjrft5ojtbr67ypjdye	139	10	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	139	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	139	12	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	139	13	found	find	VBD
work_55utqx7tjrft5ojtbr67ypjdye	139	14	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	139	15	too	too	RB
work_55utqx7tjrft5ojtbr67ypjdye	139	16	general	general	JJ
work_55utqx7tjrft5ojtbr67ypjdye	139	17	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	139	18	reference	reference	NN
work_55utqx7tjrft5ojtbr67ypjdye	139	19	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	139	20	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	139	21	semantic	semantic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	139	22	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	139	23	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	139	24	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	139	25	narrow	narrow	JJ
work_55utqx7tjrft5ojtbr67ypjdye	139	26	context	context	NN
work_55utqx7tjrft5ojtbr67ypjdye	139	27	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	139	28	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	139	29	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	139	30	sci-	sci-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	139	31	entific	entific	JJ
work_55utqx7tjrft5ojtbr67ypjdye	139	32	paper	paper	NN
work_55utqx7tjrft5ojtbr67ypjdye	139	33	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	139	34	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	139	35	actor	actor	NN
work_55utqx7tjrft5ojtbr67ypjdye	139	36	biography	biography	NN
work_55utqx7tjrft5ojtbr67ypjdye	139	37	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	140	1	4.3	4.3	LS
work_55utqx7tjrft5ojtbr67ypjdye	140	2	Clustering	Clustering	NNP
work_55utqx7tjrft5ojtbr67ypjdye	140	3	Consistency	Consistency	NNP
work_55utqx7tjrft5ojtbr67ypjdye	140	4	If	if	IN
work_55utqx7tjrft5ojtbr67ypjdye	140	5	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	140	6	treat	treat	VBP
work_55utqx7tjrft5ojtbr67ypjdye	140	7	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	140	8	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	140	9	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	140	10	clusterings	clustering	NNS
work_55utqx7tjrft5ojtbr67ypjdye	140	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	140	12	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	140	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	140	14	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	140	15	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	140	16	evaluate	evaluate	VB
work_55utqx7tjrft5ojtbr67ypjdye	140	17	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	140	18	consistent	consistent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	140	19	those	those	DT
work_55utqx7tjrft5ojtbr67ypjdye	140	20	clusters	cluster	NNS
work_55utqx7tjrft5ojtbr67ypjdye	140	21	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	140	22	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	141	1	Variation	variation	NN
work_55utqx7tjrft5ojtbr67ypjdye	141	2	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	141	3	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	141	4	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	141	5	VOI	VOI	NNP
work_55utqx7tjrft5ojtbr67ypjdye	141	6	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	141	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	141	8	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	141	9	symmetric	symmetric	JJ
work_55utqx7tjrft5ojtbr67ypjdye	141	10	mea-	mea-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	141	11	surement	surement	NN
work_55utqx7tjrft5ojtbr67ypjdye	141	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	141	13	difference	difference	NN
work_55utqx7tjrft5ojtbr67ypjdye	141	14	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	141	15	clusterings	clustering	NNS
work_55utqx7tjrft5ojtbr67ypjdye	141	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	141	17	allows	allow	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	141	18	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	141	19	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	141	20	evaluate	evaluate	VB
work_55utqx7tjrft5ojtbr67ypjdye	141	21	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	141	22	much	much	JJ
work_55utqx7tjrft5ojtbr67ypjdye	141	23	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	141	24	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	141	25	difference	difference	NN
work_55utqx7tjrft5ojtbr67ypjdye	141	26	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	141	27	makes	make	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	141	28	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	141	29	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	141	30	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	141	31	formed	form	VBN
work_55utqx7tjrft5ojtbr67ypjdye	141	32	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	141	33	Meilă	Meilă	NNP
work_55utqx7tjrft5ojtbr67ypjdye	141	34	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	141	35	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	141	36	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	141	37	Grimmer	Grimmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	141	38	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	141	39	King	King	NNP
work_55utqx7tjrft5ojtbr67ypjdye	141	40	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	141	41	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	141	42	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	141	43	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	142	1	Although	although	IN
work_55utqx7tjrft5ojtbr67ypjdye	142	2	some	some	DT
work_55utqx7tjrft5ojtbr67ypjdye	142	3	degree	degree	NN
work_55utqx7tjrft5ojtbr67ypjdye	142	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	142	5	vari-	vari-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	142	6	ation	ation	NN
work_55utqx7tjrft5ojtbr67ypjdye	142	7	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	142	8	inevitable	inevitable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	142	9	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	142	10	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	142	11	trials	trial	NNS
work_55utqx7tjrft5ojtbr67ypjdye	142	12	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	142	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	142	14	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	142	15	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	142	16	due	due	IN
work_55utqx7tjrft5ojtbr67ypjdye	142	17	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	142	18	randomness	randomness	NN
work_55utqx7tjrft5ojtbr67ypjdye	142	19	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	142	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	142	21	inference	inference	NN
work_55utqx7tjrft5ojtbr67ypjdye	142	22	algorithm	algorithm	NN
work_55utqx7tjrft5ojtbr67ypjdye	142	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	142	24	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	142	25	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	142	26	affect	affect	VB
work_55utqx7tjrft5ojtbr67ypjdye	142	27	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	142	28	much	much	JJ
work_55utqx7tjrft5ojtbr67ypjdye	142	29	occurs	occur	NNS
work_55utqx7tjrft5ojtbr67ypjdye	142	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	143	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	143	2	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	143	3	two	two	CD
work_55utqx7tjrft5ojtbr67ypjdye	143	4	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	143	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	143	6	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	143	7	metrics	metric	NNS
work_55utqx7tjrft5ojtbr67ypjdye	143	8	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	143	9	examine	examine	VB
work_55utqx7tjrft5ojtbr67ypjdye	143	10	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	143	11	stability	stability	NN
work_55utqx7tjrft5ojtbr67ypjdye	143	12	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	143	13	differences	difference	NNS
work_55utqx7tjrft5ojtbr67ypjdye	143	14	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	143	15	intra	intra	JJ
work_55utqx7tjrft5ojtbr67ypjdye	143	16	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	143	17	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	143	18	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	143	19	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	143	20	inter	inter	JJ
work_55utqx7tjrft5ojtbr67ypjdye	143	21	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	143	22	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	143	23	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	143	24	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	144	1	Intra	intra	JJ
work_55utqx7tjrft5ojtbr67ypjdye	144	2	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	144	3	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	144	4	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	144	5	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	144	6	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	144	7	be-	be-	NN
work_55utqx7tjrft5ojtbr67ypjdye	144	8	tween	tween	NN
work_55utqx7tjrft5ojtbr67ypjdye	144	9	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	144	10	trained	train	VBN
work_55utqx7tjrft5ojtbr67ypjdye	144	11	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	144	12	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	144	13	random	random	JJ
work_55utqx7tjrft5ojtbr67ypjdye	144	14	initial-	initial-	NN
work_55utqx7tjrft5ojtbr67ypjdye	144	15	izations	ization	NNS
work_55utqx7tjrft5ojtbr67ypjdye	144	16	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	144	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	144	18	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	144	19	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	144	20	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	145	1	Correspondingly	correspondingly	RB
work_55utqx7tjrft5ojtbr67ypjdye	145	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	145	3	inter	inter	JJ
work_55utqx7tjrft5ojtbr67ypjdye	145	4	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	145	5	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	145	6	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	145	7	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	145	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	145	9	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	145	10	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	145	11	outputted	outputted	JJ
work_55utqx7tjrft5ojtbr67ypjdye	145	12	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	145	13	assignments	assignment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	145	14	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	145	15	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	145	16	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	145	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	146	1	If	if	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	146	3	inter	inter	JJ
work_55utqx7tjrft5ojtbr67ypjdye	146	4	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	146	5	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	146	6	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	146	7	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	146	8	equal	equal	JJ
work_55utqx7tjrft5ojtbr67ypjdye	146	9	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	146	11	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	146	12	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	13	tri-	tri-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	146	14	als	al	NNS
work_55utqx7tjrft5ojtbr67ypjdye	146	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	146	17	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	146	18	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	146	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	146	20	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	146	21	infer	infer	VBP
work_55utqx7tjrft5ojtbr67ypjdye	146	22	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	146	24	change	change	NN
work_55utqx7tjrft5ojtbr67ypjdye	146	25	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	26	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	146	27	has	have	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	146	28	made	make	VBN
work_55utqx7tjrft5ojtbr67ypjdye	146	29	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	146	30	negligible	negligible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	146	31	difference	difference	NN
work_55utqx7tjrft5ojtbr67ypjdye	146	32	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	33	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	146	34	assignment	assignment	NN
work_55utqx7tjrft5ojtbr67ypjdye	146	35	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	36	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	146	37	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	146	38	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	146	39	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	147	1	4.4	4.4	CD
work_55utqx7tjrft5ojtbr67ypjdye	147	2	Influential	influential	JJ
work_55utqx7tjrft5ojtbr67ypjdye	147	3	Words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	147	4	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	147	5	metrics	metric	NNS
work_55utqx7tjrft5ojtbr67ypjdye	147	6	above	above	RB
work_55utqx7tjrft5ojtbr67ypjdye	147	7	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	147	8	all	all	RB
work_55utqx7tjrft5ojtbr67ypjdye	147	9	summary	summary	NN
work_55utqx7tjrft5ojtbr67ypjdye	147	10	statistics	statistic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	147	11	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	147	12	measure	measure	VBP
work_55utqx7tjrft5ojtbr67ypjdye	147	13	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	147	14	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	147	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	147	16	overall	overall	JJ
work_55utqx7tjrft5ojtbr67ypjdye	147	17	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	147	18	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	147	19	qual-	qual-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	147	20	ity	ity	NNP
work_55utqx7tjrft5ojtbr67ypjdye	147	21	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	148	1	However	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	148	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	148	3	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	148	4	understand	understand	VB
work_55utqx7tjrft5ojtbr67ypjdye	148	5	why	why	WRB
work_55utqx7tjrft5ojtbr67ypjdye	148	6	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	148	7	metrics	metric	NNS
work_55utqx7tjrft5ojtbr67ypjdye	148	8	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	148	9	affected	affect	VBN
work_55utqx7tjrft5ojtbr67ypjdye	148	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	148	11	way	way	NN
work_55utqx7tjrft5ojtbr67ypjdye	148	12	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	148	13	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	148	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	148	15	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	148	16	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	148	17	need	need	VBP
work_55utqx7tjrft5ojtbr67ypjdye	148	18	some	some	DT
work_55utqx7tjrft5ojtbr67ypjdye	148	19	way	way	NN
work_55utqx7tjrft5ojtbr67ypjdye	148	20	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	148	21	examine	examine	VB
work_55utqx7tjrft5ojtbr67ypjdye	148	22	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	148	23	individual	individual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	148	24	components	component	NNS
work_55utqx7tjrft5ojtbr67ypjdye	148	25	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	148	26	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	148	27	af-	af-	VBN
work_55utqx7tjrft5ojtbr67ypjdye	148	28	fected	fecte	VBN
work_55utqx7tjrft5ojtbr67ypjdye	148	29	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	148	30	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	148	31	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	148	32	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	148	33	available	available	JJ
work_55utqx7tjrft5ojtbr67ypjdye	148	34	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	148	35	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	148	36	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	148	37	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	149	1	291	291	CD
work_55utqx7tjrft5ojtbr67ypjdye	149	2	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	149	3	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	149	4	two	two	CD
work_55utqx7tjrft5ojtbr67ypjdye	149	5	heuristics	heuristic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	149	6	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	149	7	identify	identify	VB
work_55utqx7tjrft5ojtbr67ypjdye	149	8	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	149	9	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	149	10	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	149	11	most	most	RBS
work_55utqx7tjrft5ojtbr67ypjdye	149	12	affected	affect	VBN
work_55utqx7tjrft5ojtbr67ypjdye	149	13	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	149	14	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	149	15	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	149	16	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	149	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	150	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	150	2	first	first	JJ
work_55utqx7tjrft5ojtbr67ypjdye	150	3	uses	use	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	150	4	inferred	inferred	JJ
work_55utqx7tjrft5ojtbr67ypjdye	150	5	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	150	6	probabilities	probability	NNS
work_55utqx7tjrft5ojtbr67ypjdye	150	7	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	150	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	150	9	test	test	NN
work_55utqx7tjrft5ojtbr67ypjdye	150	10	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	150	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	151	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	151	2	want	want	VBP
work_55utqx7tjrft5ojtbr67ypjdye	151	3	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	151	4	scoring	score	VBG
work_55utqx7tjrft5ojtbr67ypjdye	151	5	function	function	NN
work_55utqx7tjrft5ojtbr67ypjdye	151	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	151	7	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	151	8	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	151	9	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	151	10	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	151	11	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	151	12	positive	positive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	151	13	if	if	IN
work_55utqx7tjrft5ojtbr67ypjdye	151	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	151	15	estimated	estimate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	151	16	joint	joint	JJ
work_55utqx7tjrft5ojtbr67ypjdye	151	17	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	151	18	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	151	19	to-	to-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	151	20	kens	ken	NNS
work_55utqx7tjrft5ojtbr67ypjdye	151	21	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	151	22	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	151	23	particular	particular	JJ
work_55utqx7tjrft5ojtbr67ypjdye	151	24	pre	pre	JJ
work_55utqx7tjrft5ojtbr67ypjdye	151	25	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	151	26	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	151	27	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	151	28	increases	increase	NNS
work_55utqx7tjrft5ojtbr67ypjdye	151	29	af-	af-	IN
work_55utqx7tjrft5ojtbr67ypjdye	151	30	ter	ter	NN
work_55utqx7tjrft5ojtbr67ypjdye	151	31	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	151	32	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	151	33	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	151	34	negative	negative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	151	35	if	if	IN
work_55utqx7tjrft5ojtbr67ypjdye	151	36	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	151	37	decreases	decrease	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	151	38	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	152	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	152	2	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	152	3	want	want	VBP
work_55utqx7tjrft5ojtbr67ypjdye	152	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	152	5	magnitude	magnitude	NN
work_55utqx7tjrft5ojtbr67ypjdye	152	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	152	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	152	8	score	score	NN
work_55utqx7tjrft5ojtbr67ypjdye	152	9	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	152	10	correspond	correspond	VB
work_55utqx7tjrft5ojtbr67ypjdye	152	11	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	152	12	both	both	CC
work_55utqx7tjrft5ojtbr67ypjdye	152	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	152	14	difference	difference	NN
work_55utqx7tjrft5ojtbr67ypjdye	152	15	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	152	16	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	152	17	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	152	18	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	152	19	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	152	20	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	152	21	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	152	22	relative	relative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	152	23	informativeness	informativeness	NN
work_55utqx7tjrft5ojtbr67ypjdye	152	24	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	152	25	that	that	DT
work_55utqx7tjrft5ojtbr67ypjdye	152	26	token	token	VBN
work_55utqx7tjrft5ojtbr67ypjdye	152	27	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	152	28	dis-	dis-	IN
work_55utqx7tjrft5ojtbr67ypjdye	152	29	tinguishing	tinguishe	VBG
work_55utqx7tjrft5ojtbr67ypjdye	152	30	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	152	31	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	152	32	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	152	33	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	153	1	For	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	153	2	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	153	3	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	153	4	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	5	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	6	w	w	NNP
work_55utqx7tjrft5ojtbr67ypjdye	153	7	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	153	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	153	9	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	153	10	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	153	12	function	function	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	13	t	t	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	14	applying	apply	VBG
work_55utqx7tjrft5ojtbr67ypjdye	153	15	some	some	DT
work_55utqx7tjrft5ojtbr67ypjdye	153	16	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	17	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	153	19	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	153	20	compute	compute	VBP
work_55utqx7tjrft5ojtbr67ypjdye	153	21	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	153	22	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	23	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	24	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	25	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	153	26	TPwt	TPwt	NNP
work_55utqx7tjrft5ojtbr67ypjdye	153	27	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	153	28	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	153	29	D∑	D∑	NNP
work_55utqx7tjrft5ojtbr67ypjdye	153	30	d=1	d=1	NN
work_55utqx7tjrft5ojtbr67ypjdye	153	31	Nd∑	Nd∑	NNP
work_55utqx7tjrft5ojtbr67ypjdye	153	32	i=1	i=1	NNP
work_55utqx7tjrft5ojtbr67ypjdye	153	33	I[xdi	I[xdi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	153	34	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	153	35	w	w	NNP
work_55utqx7tjrft5ojtbr67ypjdye	153	36	]	]	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	153	37	log(P(t(xdi	log(P(t(xdi	.
work_55utqx7tjrft5ojtbr67ypjdye	153	38	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	153	39	=	=	NFP
work_55utqx7tjrft5ojtbr67ypjdye	153	40	t(w)|	t(w)|	CD
work_55utqx7tjrft5ojtbr67ypjdye	153	41	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	154	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	155	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	156	1	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	156	2	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	156	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	156	4	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	156	5	5	5	CD
work_55utqx7tjrft5ojtbr67ypjdye	156	6	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	156	7	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	156	8	D	d	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	9	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	156	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	156	11	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	156	13	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	156	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	156	15	Nd	Nd	NNP
work_55utqx7tjrft5ojtbr67ypjdye	156	16	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	156	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	156	18	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	19	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	156	20	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	156	21	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	156	22	document	document	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	23	d	d	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	24	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	156	25	xdi	xdi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	156	26	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	156	27	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	156	28	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	156	29	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	30	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	31	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	156	32	token	token	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	33	i	i	PRP
work_55utqx7tjrft5ojtbr67ypjdye	156	34	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	156	35	document	document	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	36	d	d	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	37	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	156	38	t(xdi	t(xdi	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	39	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	156	40	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	156	41	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	156	42	treated	treat	VBN
work_55utqx7tjrft5ojtbr67ypjdye	156	43	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	44	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	156	45	I[xdi	I[xdi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	156	46	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	156	47	w	w	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	48	]	]	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	156	49	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	156	50	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	156	51	indicator	indicator	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	52	func-	func-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	156	53	tion	tion	NN
work_55utqx7tjrft5ojtbr67ypjdye	156	54	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	156	55	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	156	56	1	1	CD
work_55utqx7tjrft5ojtbr67ypjdye	156	57	if	if	IN
work_55utqx7tjrft5ojtbr67ypjdye	156	58	xdi	xdi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	156	59	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	156	60	w	w	NNP
work_55utqx7tjrft5ojtbr67ypjdye	156	61	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	156	62	zero	zero	CD
work_55utqx7tjrft5ojtbr67ypjdye	156	63	otherwise	otherwise	RB
work_55utqx7tjrft5ojtbr67ypjdye	156	64	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	156	65	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	156	66	P(t(xdi	P(t(xdi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	156	67	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	156	68	=	=	NFP
work_55utqx7tjrft5ojtbr67ypjdye	156	69	t(w)|	t(w)|	CD
work_55utqx7tjrft5ojtbr67ypjdye	156	70	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	157	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	158	1	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	158	2	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	159	1	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	159	2	shorthand	shorthand	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	3	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	159	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	159	5	held-	held-	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	6	out	out	RP
work_55utqx7tjrft5ojtbr67ypjdye	159	7	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	8	estimate	estimate	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	159	10	treated	treat	VBN
work_55utqx7tjrft5ojtbr67ypjdye	159	11	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	159	12	t(xdi	t(xdi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	159	13	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	159	14	hav-	hav-	NNS
work_55utqx7tjrft5ojtbr67ypjdye	159	15	ing	e	VBG
work_55utqx7tjrft5ojtbr67ypjdye	159	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	159	17	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	18	w	w	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	19	generated	generate	VBD
work_55utqx7tjrft5ojtbr67ypjdye	159	20	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	159	21	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	159	22	left	left	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	23	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	159	24	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	159	25	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	159	26	right	right	JJ
work_55utqx7tjrft5ojtbr67ypjdye	159	27	tech-	tech-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	159	28	nique	nique	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	29	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	159	30	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	159	31	inferred	inferred	JJ
work_55utqx7tjrft5ojtbr67ypjdye	159	32	parameters	parameter	NNS
work_55utqx7tjrft5ojtbr67ypjdye	159	33	θ	θ	NNP
work_55utqx7tjrft5ojtbr67ypjdye	159	34	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	159	35	φ	φ	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	36	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	159	37	hyper-	hyper-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	159	38	parameters	parameter	NNS
work_55utqx7tjrft5ojtbr67ypjdye	159	39	α	α	XX
work_55utqx7tjrft5ojtbr67ypjdye	159	40	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	159	41	β	β	NNP
work_55utqx7tjrft5ojtbr67ypjdye	159	42	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	159	43	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	159	44	trained	train	VBN
work_55utqx7tjrft5ojtbr67ypjdye	159	45	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	159	46	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	160	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	160	2	average	average	VBP
work_55utqx7tjrft5ojtbr67ypjdye	160	3	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	160	4	quantity	quantity	NN
work_55utqx7tjrft5ojtbr67ypjdye	160	5	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	160	6	Equation	Equation	NNP
work_55utqx7tjrft5ojtbr67ypjdye	160	7	5	5	CD
work_55utqx7tjrft5ojtbr67ypjdye	160	8	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	160	9	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	160	10	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	160	11	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	160	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	160	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	160	14	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	160	15	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	160	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	160	17	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	160	18	count	count	NN
work_55utqx7tjrft5ojtbr67ypjdye	160	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	160	20	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	160	21	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	160	22	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	160	23	get	get	VB
work_55utqx7tjrft5ojtbr67ypjdye	160	24	TPwt	TPwt	NNP
work_55utqx7tjrft5ojtbr67ypjdye	160	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	161	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	161	2	order	order	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	3	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	161	4	compute	compute	VB
work_55utqx7tjrft5ojtbr67ypjdye	161	5	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	161	6	rela-	rela-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	161	7	tive	tive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	161	8	score	score	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	161	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	161	11	amount	amount	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	161	13	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	14	improvement	improvement	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	161	16	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	161	17	individual	individual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	161	18	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	19	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	161	20	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	161	21	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	22	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	23	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	161	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	161	25	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	161	26	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	161	27	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	28	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	29	t0	t0	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	30	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	161	31	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	161	32	take	take	VBP
work_55utqx7tjrft5ojtbr67ypjdye	161	33	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	161	34	difference	difference	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	35	be-	be-	XX
work_55utqx7tjrft5ojtbr67ypjdye	161	36	tween	tween	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	37	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	161	38	probabilities	probability	NNS
work_55utqx7tjrft5ojtbr67ypjdye	161	39	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	161	40	weighted	weight	VBN
work_55utqx7tjrft5ojtbr67ypjdye	161	41	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	161	42	inverse	inverse	NNP
work_55utqx7tjrft5ojtbr67ypjdye	161	43	docu-	docu-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	161	44	ment	ment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	161	45	frequency	frequency	NNP
work_55utqx7tjrft5ojtbr67ypjdye	161	46	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	161	47	idf	idf	NNP
work_55utqx7tjrft5ojtbr67ypjdye	161	48	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	161	49	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	161	50	favor	favor	VB
work_55utqx7tjrft5ojtbr67ypjdye	161	51	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	161	52	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	161	53	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	161	54	specific	specific	JJ
work_55utqx7tjrft5ojtbr67ypjdye	161	55	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	161	56	particular	particular	JJ
work_55utqx7tjrft5ojtbr67ypjdye	161	57	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	161	58	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	162	1	Our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	162	2	final	final	JJ
work_55utqx7tjrft5ojtbr67ypjdye	162	3	score	score	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	4	function	function	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	5	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	162	6	TPscorewt	TPscorewt	NNP
work_55utqx7tjrft5ojtbr67ypjdye	162	7	=	=	NFP
work_55utqx7tjrft5ojtbr67ypjdye	162	8	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	162	9	TPwt	TPwt	NNP
work_55utqx7tjrft5ojtbr67ypjdye	162	10	−TPwt0	−TPwt0	NNP
work_55utqx7tjrft5ojtbr67ypjdye	162	11	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	162	12	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	13	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	162	14	D	D	NNP
work_55utqx7tjrft5ojtbr67ypjdye	162	15	Dw	Dw	NNP
work_55utqx7tjrft5ojtbr67ypjdye	162	16	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	162	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	162	18	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	162	19	6	6	CD
work_55utqx7tjrft5ojtbr67ypjdye	162	20	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	162	21	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	162	22	Dw	Dw	NNP
work_55utqx7tjrft5ojtbr67ypjdye	162	23	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	162	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	162	25	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	26	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	162	27	documents	document	NNS
work_55utqx7tjrft5ojtbr67ypjdye	162	28	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	162	29	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	162	30	total	total	JJ
work_55utqx7tjrft5ojtbr67ypjdye	162	31	D	d	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	32	containing	contain	VBG
work_55utqx7tjrft5ojtbr67ypjdye	162	33	at	at	RB
work_55utqx7tjrft5ojtbr67ypjdye	162	34	least	least	RBS
work_55utqx7tjrft5ojtbr67ypjdye	162	35	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	162	36	token	token	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	37	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	162	38	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	39	w.	w.	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	40	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	162	41	lowest	low	JJS
work_55utqx7tjrft5ojtbr67ypjdye	162	42	negative	negative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	162	43	scores	score	NNS
work_55utqx7tjrft5ojtbr67ypjdye	162	44	indicate	indicate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	162	45	higher	high	JJR
work_55utqx7tjrft5ojtbr67ypjdye	162	46	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	47	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	162	48	im-	im-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	162	49	portance	portance	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	50	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	162	51	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	162	52	unstemmed	unstemmed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	162	53	form	form	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	54	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	162	55	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	162	56	token	token	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	57	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	162	58	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	162	59	high	high	JJ
work_55utqx7tjrft5ojtbr67ypjdye	162	60	positive	positive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	162	61	scores	score	NNS
work_55utqx7tjrft5ojtbr67ypjdye	162	62	indicate	indicate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	162	63	higher	high	JJR
work_55utqx7tjrft5ojtbr67ypjdye	162	64	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	65	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	162	66	importance	importance	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	67	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	162	68	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	162	69	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	162	70	form	form	NN
work_55utqx7tjrft5ojtbr67ypjdye	162	71	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	163	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	2	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	163	3	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	163	4	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	163	5	produce	produce	VB
work_55utqx7tjrft5ojtbr67ypjdye	163	6	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	163	7	symmetric	symmetric	JJ
work_55utqx7tjrft5ojtbr67ypjdye	163	8	distribution	distribution	NN
work_55utqx7tjrft5ojtbr67ypjdye	163	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	163	10	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	11	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	163	12	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	163	13	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	163	14	accounted	account	VBN
work_55utqx7tjrft5ojtbr67ypjdye	163	15	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	163	17	increased	increase	VBN
work_55utqx7tjrft5ojtbr67ypjdye	163	18	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	163	19	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	20	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	163	21	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	163	22	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	23	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	163	24	smaller	small	JJR
work_55utqx7tjrft5ojtbr67ypjdye	163	25	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	163	26	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	163	27	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	163	28	allows	allow	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	163	29	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	163	30	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	163	31	sort	sort	VB
work_55utqx7tjrft5ojtbr67ypjdye	163	32	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	163	33	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	34	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	163	35	much	much	JJ
work_55utqx7tjrft5ojtbr67ypjdye	163	36	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	163	37	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	163	38	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	39	occurring	occur	VBG
work_55utqx7tjrft5ojtbr67ypjdye	163	40	has	have	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	163	41	changed	change	VBN
work_55utqx7tjrft5ojtbr67ypjdye	163	42	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	43	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	163	44	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	163	45	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	163	46	much	much	JJ
work_55utqx7tjrft5ojtbr67ypjdye	163	47	that	that	DT
work_55utqx7tjrft5ojtbr67ypjdye	163	48	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	163	49	affects	affect	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	163	50	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	163	51	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	163	52	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	163	53	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	163	54	whole	whole	NN
work_55utqx7tjrft5ojtbr67ypjdye	163	55	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	164	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	164	2	second	second	JJ
work_55utqx7tjrft5ojtbr67ypjdye	164	3	heuristic	heuristic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	164	4	tests	test	NNS
work_55utqx7tjrft5ojtbr67ypjdye	164	5	whether	whether	IN
work_55utqx7tjrft5ojtbr67ypjdye	164	6	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	164	7	in-	in-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	164	8	creases	crease	NNS
work_55utqx7tjrft5ojtbr67ypjdye	164	9	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	164	10	decreases	decrease	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	164	11	certainty	certainty	NN
work_55utqx7tjrft5ojtbr67ypjdye	164	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	164	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	164	14	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	164	15	assign-	assign-	NN
work_55utqx7tjrft5ojtbr67ypjdye	164	16	ment	ment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	164	17	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	164	18	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	164	19	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	164	20	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	164	21	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	164	22	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	165	1	Intuitively	intuitively	RB
work_55utqx7tjrft5ojtbr67ypjdye	165	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	165	3	cor-	cor-	RB
work_55utqx7tjrft5ojtbr67ypjdye	165	4	rect	rect	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	5	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	6	should	should	MD
work_55utqx7tjrft5ojtbr67ypjdye	165	7	reduce	reduce	VB
work_55utqx7tjrft5ojtbr67ypjdye	165	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	165	9	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	10	en-	en-	RB
work_55utqx7tjrft5ojtbr67ypjdye	165	11	tropy	tropy	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	12	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	165	13	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	165	14	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	165	15	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	165	16	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	165	17	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	18	class	class	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	19	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	165	20	forcing	force	VBG
work_55utqx7tjrft5ojtbr67ypjdye	165	21	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	165	22	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	165	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	165	24	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	165	25	root	root	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	26	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	165	27	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	165	28	treated	treat	VBN
work_55utqx7tjrft5ojtbr67ypjdye	165	29	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	165	30	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	165	31	single	single	JJ
work_55utqx7tjrft5ojtbr67ypjdye	165	32	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	33	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	165	34	inference	inference	NN
work_55utqx7tjrft5ojtbr67ypjdye	165	35	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	166	1	Topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	166	2	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	166	3	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	166	4	in-	in-	DT
work_55utqx7tjrft5ojtbr67ypjdye	166	5	tuitively	tuitively	RB
work_55utqx7tjrft5ojtbr67ypjdye	166	6	better	well	RBR
work_55utqx7tjrft5ojtbr67ypjdye	166	7	when	when	WRB
work_55utqx7tjrft5ojtbr67ypjdye	166	8	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	166	9	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	166	10	sparsely	sparsely	RB
work_55utqx7tjrft5ojtbr67ypjdye	166	11	distributed	distribute	VBN
work_55utqx7tjrft5ojtbr67ypjdye	166	12	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	166	13	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	166	14	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	166	15	consequently	consequently	RB
work_55utqx7tjrft5ojtbr67ypjdye	166	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	166	17	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	166	18	prefer	prefer	VBP
work_55utqx7tjrft5ojtbr67ypjdye	166	19	lower	lower	RBR
work_55utqx7tjrft5ojtbr67ypjdye	166	20	entropy	entropy	RB
work_55utqx7tjrft5ojtbr67ypjdye	166	21	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	166	22	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	166	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	166	24	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	166	25	mass	mass	NN
work_55utqx7tjrft5ojtbr67ypjdye	166	26	concentrated	concentrate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	166	27	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	166	28	only	only	RB
work_55utqx7tjrft5ojtbr67ypjdye	166	29	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	166	30	few	few	JJ
work_55utqx7tjrft5ojtbr67ypjdye	166	31	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	166	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	167	1	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	167	2	negative	negative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	167	3	value	value	NN
work_55utqx7tjrft5ojtbr67ypjdye	167	4	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	167	5	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	167	6	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	167	7	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	167	8	under	under	IN
work_55utqx7tjrft5ojtbr67ypjdye	167	9	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	167	10	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	167	11	metric	metric	JJ
work_55utqx7tjrft5ojtbr67ypjdye	167	12	favors	favor	NNS
work_55utqx7tjrft5ojtbr67ypjdye	167	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	167	14	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	167	15	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	167	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	167	17	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	167	18	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	167	19	positive	positive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	167	20	score	score	NN
work_55utqx7tjrft5ojtbr67ypjdye	167	21	favors	favor	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	167	22	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	167	23	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	167	24	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	167	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	168	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	2	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	3	case	case	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	5	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	6	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	7	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	168	8	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	9	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	10	w	w	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	12	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	168	13	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	168	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	15	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	16	assignments	assignment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	17	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	18	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	19	final	final	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	20	iteration	iteration	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	21	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	22	Gibbs	Gibbs	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	168	23	sampling	sample	VBG
work_55utqx7tjrft5ojtbr67ypjdye	168	24	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	168	25	compute	compute	VB
work_55utqx7tjrft5ojtbr67ypjdye	168	26	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	27	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	28	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	29	instances	instance	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	30	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	31	w	w	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	32	assigned	assign	VBN
work_55utqx7tjrft5ojtbr67ypjdye	168	33	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	34	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	35	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	36	k.	k.	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	37	To	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	168	38	preserve	preserve	VB
work_55utqx7tjrft5ojtbr67ypjdye	168	39	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	40	sparsity	sparsity	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	41	inferred	infer	VBN
work_55utqx7tjrft5ojtbr67ypjdye	168	42	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	43	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	44	algorithm	algorithm	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	45	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	46	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	168	47	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	168	48	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	49	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	168	50	generate	generate	VB
work_55utqx7tjrft5ojtbr67ypjdye	168	51	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	52	maximum	maximum	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	53	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	168	54	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	55	estimate	estimate	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	56	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	57	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	58	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	59	dis-	dis-	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	60	tribution	tribution	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	61	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	62	w	w	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	63	being	be	VBG
work_55utqx7tjrft5ojtbr67ypjdye	168	64	assigned	assign	VBN
work_55utqx7tjrft5ojtbr67ypjdye	168	65	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	66	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	67	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	68	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	69	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	70	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	168	71	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	168	72	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	168	73	compute	compute	VB
work_55utqx7tjrft5ojtbr67ypjdye	168	74	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	75	Shannon	Shannon	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	76	entropy	entropy	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	77	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	168	78	Hwt(k	hwt(k	LS
work_55utqx7tjrft5ojtbr67ypjdye	168	79	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	80	=	=	NFP
work_55utqx7tjrft5ojtbr67ypjdye	168	81	−	−	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	82	K∑	K∑	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	83	k=1	k=1	VBD
work_55utqx7tjrft5ojtbr67ypjdye	168	84	Nwk	Nwk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	85	Nw	Nw	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	86	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	87	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	88	Nwk	Nwk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	89	Nw	Nw	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	90	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	91	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	92	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	93	7	7	LS
work_55utqx7tjrft5ojtbr67ypjdye	168	94	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	95	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	168	96	Nwk	Nwk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	97	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	168	98	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	99	count	count	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	100	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	101	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	102	tokens	token	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	103	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	104	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	105	w	w	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	106	as-	as-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	107	signed	sign	VBD
work_55utqx7tjrft5ojtbr67ypjdye	168	108	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	109	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	110	k.	k.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	111	For	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	112	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	113	treated	treat	VBN
work_55utqx7tjrft5ojtbr67ypjdye	168	114	form	form	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	115	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	116	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	117	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	118	w	w	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	119	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	120	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	121	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	122	t	t	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	123	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	124	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	168	125	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	168	126	consider	consider	VBP
work_55utqx7tjrft5ojtbr67ypjdye	168	127	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	128	inverse	inverse	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	129	im-	im-	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	130	age	age	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	131	t−1(w	t−1(w	CD
work_55utqx7tjrft5ojtbr67ypjdye	168	132	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	133	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	134	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	168	135	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	136	set	set	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	137	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	138	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	139	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	140	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	168	141	stem	stem	VBP
work_55utqx7tjrft5ojtbr67ypjdye	168	142	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	168	143	have	have	VB
work_55utqx7tjrft5ojtbr67ypjdye	168	144	form	form	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	145	w.	w.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	146	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	168	147	therefore	therefore	RB
work_55utqx7tjrft5ojtbr67ypjdye	168	148	compute	compute	VBP
work_55utqx7tjrft5ojtbr67ypjdye	168	149	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	150	change	change	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	151	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	152	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	153	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	168	154	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	155	H̄wt	H̄wt	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	156	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	157	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	158	trials	trial	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	159	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	160	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	161	t	t	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	162	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	168	163	control	control	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	164	t0	t0	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	165	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	166	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	167	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	168	168	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	169	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	168	170	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	171	count	count	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	172	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	173	∆Hwt(k	∆hwt(k	ADD
work_55utqx7tjrft5ojtbr67ypjdye	168	174	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	175	=	=	NFP
work_55utqx7tjrft5ojtbr67ypjdye	168	176	H̄wt(k	h̄wt(k	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	177	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	178	−	−	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	179	H̄t−1(w)t0	H̄t−1(w)t0	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	180	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	181	k	k	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	182	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	183	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	184	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	168	185	8)	8)	CD
work_55utqx7tjrft5ojtbr67ypjdye	168	186	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	168	187	H̄t−1(w)t0	H̄t−1(w)t0	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	188	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	168	189	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	190	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	191	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	192	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	193	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	194	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	195	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	168	196	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	197	counts	count	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	198	summed	sum	VBD
work_55utqx7tjrft5ojtbr67ypjdye	168	199	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	200	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	201	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	202	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	203	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	168	204	conflate	conflate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	168	205	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	168	206	type	type	VB
work_55utqx7tjrft5ojtbr67ypjdye	168	207	w	w	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	208	under	under	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	209	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	210	t.	t.	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	211	5	5	CD
work_55utqx7tjrft5ojtbr67ypjdye	168	212	Results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	213	To	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	168	214	evaluate	evaluate	VB
work_55utqx7tjrft5ojtbr67ypjdye	168	215	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	168	216	effects	effect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	217	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	218	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	219	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	168	220	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	221	mod-	mod-	NN
work_55utqx7tjrft5ojtbr67ypjdye	168	222	els	els	NNP
work_55utqx7tjrft5ojtbr67ypjdye	168	223	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	168	224	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	168	225	produced	produce	VBD
work_55utqx7tjrft5ojtbr67ypjdye	168	226	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	227	thousand	thousand	CD
work_55utqx7tjrft5ojtbr67ypjdye	168	228	inferred	inferred	JJ
work_55utqx7tjrft5ojtbr67ypjdye	168	229	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	168	230	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	169	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	169	2	apply	apply	VBP
work_55utqx7tjrft5ojtbr67ypjdye	169	3	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	169	4	metrics	metric	NNS
work_55utqx7tjrft5ojtbr67ypjdye	169	5	described	describe	VBN
work_55utqx7tjrft5ojtbr67ypjdye	169	6	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	169	7	Section	section	NN
work_55utqx7tjrft5ojtbr67ypjdye	169	8	4	4	CD
work_55utqx7tjrft5ojtbr67ypjdye	169	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	169	10	com-	com-	NN
work_55utqx7tjrft5ojtbr67ypjdye	169	11	puting	puting	NN
work_55utqx7tjrft5ojtbr67ypjdye	169	12	means	mean	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	169	13	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	169	14	standard	standard	JJ
work_55utqx7tjrft5ojtbr67ypjdye	169	15	errors	error	NNS
work_55utqx7tjrft5ojtbr67ypjdye	169	16	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	169	17	trials	trial	NNS
work_55utqx7tjrft5ojtbr67ypjdye	169	18	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	169	19	292	292	CD
work_55utqx7tjrft5ojtbr67ypjdye	169	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	169	21	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	169	22	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	169	23	count	count	NN
work_55utqx7tjrft5ojtbr67ypjdye	169	24	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	169	25	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	169	26	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	169	27	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	169	28	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	169	29	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	169	30	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	169	31	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	169	32	ensure	ensure	VB
work_55utqx7tjrft5ojtbr67ypjdye	169	33	significance	significance	NN
work_55utqx7tjrft5ojtbr67ypjdye	169	34	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	170	1	5.1	5.1	CD
work_55utqx7tjrft5ojtbr67ypjdye	170	2	Treatment	Treatment	NNP
work_55utqx7tjrft5ojtbr67ypjdye	170	3	Strength	Strength	NNP
work_55utqx7tjrft5ojtbr67ypjdye	170	4	Many	many	JJ
work_55utqx7tjrft5ojtbr67ypjdye	170	5	factors	factor	NNS
work_55utqx7tjrft5ojtbr67ypjdye	170	6	contribute	contribute	VBP
work_55utqx7tjrft5ojtbr67ypjdye	170	7	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	170	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	170	9	general	general	JJ
work_55utqx7tjrft5ojtbr67ypjdye	170	10	concept	concept	NN
work_55utqx7tjrft5ojtbr67ypjdye	170	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	170	12	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	170	13	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	170	14	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	170	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	170	16	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	170	17	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	170	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	170	19	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	170	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	170	21	most	most	RBS
work_55utqx7tjrft5ojtbr67ypjdye	170	22	obvious	obvious	JJ
work_55utqx7tjrft5ojtbr67ypjdye	170	23	sig-	sig-	NN
work_55utqx7tjrft5ojtbr67ypjdye	170	24	nal	nal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	170	25	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	170	26	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	170	27	amount	amount	NN
work_55utqx7tjrft5ojtbr67ypjdye	170	28	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	170	29	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	170	30	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	170	31	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	170	32	reduces	reduce	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	170	33	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	170	34	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	170	35	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	170	36	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	170	37	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	170	38	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	171	1	After	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	171	2	stopword	stopword	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	3	removal	removal	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	171	5	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	171	6	count	count	VBP
work_55utqx7tjrft5ojtbr67ypjdye	171	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	171	8	total	total	JJ
work_55utqx7tjrft5ojtbr67ypjdye	171	9	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	10	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	171	11	unique	unique	JJ
work_55utqx7tjrft5ojtbr67ypjdye	171	12	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	13	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	171	14	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	171	15	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	171	16	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	171	17	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	18	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	171	19	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	171	20	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	21	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	171	22	training	training	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	23	cor-	cor-	XX
work_55utqx7tjrft5ojtbr67ypjdye	171	24	pus	pus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	171	25	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	171	26	as	as	RB
work_55utqx7tjrft5ojtbr67ypjdye	171	27	well	well	RB
work_55utqx7tjrft5ojtbr67ypjdye	171	28	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	171	29	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	171	30	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	171	31	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	32	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	171	33	characters	character	NNS
work_55utqx7tjrft5ojtbr67ypjdye	171	34	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	171	35	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	171	36	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	37	after	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	171	38	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	171	39	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	172	1	Comparing	compare	VBG
work_55utqx7tjrft5ojtbr67ypjdye	172	2	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	172	3	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	172	4	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	172	5	ratios	ratio	NNS
work_55utqx7tjrft5ojtbr67ypjdye	172	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	172	7	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	172	8	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	172	9	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	172	10	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	172	11	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	172	12	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	172	13	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	172	14	cor-	cor-	NN
work_55utqx7tjrft5ojtbr67ypjdye	172	15	pus	pus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	172	16	gives	give	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	172	17	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	172	18	measurement	measurement	NN
work_55utqx7tjrft5ojtbr67ypjdye	172	19	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	172	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	172	21	ratio	ratio	NN
work_55utqx7tjrft5ojtbr67ypjdye	172	22	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	172	23	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	172	24	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	172	25	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	172	26	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	172	27	classes	class	NNS
work_55utqx7tjrft5ojtbr67ypjdye	172	28	under	under	IN
work_55utqx7tjrft5ojtbr67ypjdye	172	29	that	that	DT
work_55utqx7tjrft5ojtbr67ypjdye	172	30	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	172	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	173	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	173	2	display	display	VBP
work_55utqx7tjrft5ojtbr67ypjdye	173	3	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	173	4	counts	count	NNS
work_55utqx7tjrft5ojtbr67ypjdye	173	5	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	173	6	Figure	Figure	NNP
work_55utqx7tjrft5ojtbr67ypjdye	173	7	1	1	CD
work_55utqx7tjrft5ojtbr67ypjdye	173	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	174	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	174	2	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	174	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	174	4	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	174	5	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	174	6	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	174	7	already	already	RB
work_55utqx7tjrft5ojtbr67ypjdye	174	8	demonstrate	demonstrate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	174	9	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	174	10	stemmer	stemmer	JJ
work_55utqx7tjrft5ojtbr67ypjdye	174	11	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	174	12	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	174	13	depend	depend	VB
work_55utqx7tjrft5ojtbr67ypjdye	174	14	heav-	heav-	NN
work_55utqx7tjrft5ojtbr67ypjdye	174	15	ily	ily	NNP
work_55utqx7tjrft5ojtbr67ypjdye	174	16	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	174	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	174	18	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	174	19	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	174	20	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	174	21	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	174	22	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	174	23	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	174	24	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	174	25	applied	apply	VBN
work_55utqx7tjrft5ojtbr67ypjdye	174	26	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	175	1	For	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	175	2	instance	instance	NN
work_55utqx7tjrft5ojtbr67ypjdye	175	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	175	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	175	5	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	175	6	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	175	7	actually	actually	RB
work_55utqx7tjrft5ojtbr67ypjdye	175	8	increases	increase	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	175	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	175	10	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	175	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	175	12	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	175	13	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	175	14	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	175	15	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	175	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	175	17	whereas	whereas	IN
work_55utqx7tjrft5ojtbr67ypjdye	175	18	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	175	19	produces	produce	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	175	20	more	more	JJR
work_55utqx7tjrft5ojtbr67ypjdye	175	21	vocabulary	vocabulary	JJ
work_55utqx7tjrft5ojtbr67ypjdye	175	22	reduction	reduction	NN
work_55utqx7tjrft5ojtbr67ypjdye	175	23	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	175	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	175	25	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	175	26	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	175	27	both	both	CC
work_55utqx7tjrft5ojtbr67ypjdye	175	28	IMDb	IMDb	NNS
work_55utqx7tjrft5ojtbr67ypjdye	175	29	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	175	30	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	175	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	176	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	176	2	proportions	proportion	NNS
work_55utqx7tjrft5ojtbr67ypjdye	176	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	176	4	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	176	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	176	6	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	176	7	stemmer	stemmer	JJ
work_55utqx7tjrft5ojtbr67ypjdye	176	8	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	176	9	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	176	10	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	176	11	ratios	ratio	NNS
work_55utqx7tjrft5ojtbr67ypjdye	176	12	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	176	13	consistent	consistent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	176	14	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	176	15	cor-	cor-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	176	16	pora	pora	NNP
work_55utqx7tjrft5ojtbr67ypjdye	176	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	176	18	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	176	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	176	20	exception	exception	NN
work_55utqx7tjrft5ojtbr67ypjdye	176	21	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	176	22	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	176	23	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	176	24	arXiv	arXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	176	25	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	177	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	177	2	frequent	frequent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	177	3	use	use	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	5	scientific	scientific	JJ
work_55utqx7tjrft5ojtbr67ypjdye	177	6	prefixes	prefix	NNS
work_55utqx7tjrft5ojtbr67ypjdye	177	7	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	177	8	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	177	9	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	10	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	177	11	inter	inter	JJ
work_55utqx7tjrft5ojtbr67ypjdye	177	12	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	177	13	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	177	14	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	177	15	anti	anti	FW
work_55utqx7tjrft5ojtbr67ypjdye	177	16	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	177	17	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	177	18	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	177	19	bad	bad	JJ
work_55utqx7tjrft5ojtbr67ypjdye	177	20	conversion	conversion	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	21	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	22	PDF	PDF	NNP
work_55utqx7tjrft5ojtbr67ypjdye	177	23	format	format	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	24	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	25	arXiv	arXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	177	26	lead	lead	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	27	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	28	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	177	29	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	177	30	conflate	conflate	VB
work_55utqx7tjrft5ojtbr67ypjdye	177	31	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	32	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	177	33	higher	high	JJR
work_55utqx7tjrft5ojtbr67ypjdye	177	34	rate	rate	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	35	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	36	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	177	37	do	do	VBP
work_55utqx7tjrft5ojtbr67ypjdye	177	38	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	39	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	177	40	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	41	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	42	re-	re-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	177	43	spect	spect	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	44	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	177	45	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	177	46	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	177	47	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	177	48	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	177	49	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	177	50	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	178	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	178	2	three	three	CD
work_55utqx7tjrft5ojtbr67ypjdye	178	3	dif-	dif-	CC
work_55utqx7tjrft5ojtbr67ypjdye	178	4	ferent	ferent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	178	5	light	light	NN
work_55utqx7tjrft5ojtbr67ypjdye	178	6	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	178	7	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	178	8	—	—	:
work_55utqx7tjrft5ojtbr67ypjdye	178	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	178	10	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	178	11	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	178	12	stemmer	stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	178	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	178	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	178	15	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	178	16	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	178	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	178	18	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	178	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	178	20	WordNet	WordNet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	178	21	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	178	22	—	—	:
work_55utqx7tjrft5ojtbr67ypjdye	178	23	perform	perform	VB
work_55utqx7tjrft5ojtbr67ypjdye	178	24	similarly	similarly	RB
work_55utqx7tjrft5ojtbr67ypjdye	178	25	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	178	26	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	178	27	IMDb	IMDb	NNP
work_55utqx7tjrft5ojtbr67ypjdye	178	28	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	178	29	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	178	30	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	178	31	vary	vary	VBP
work_55utqx7tjrft5ojtbr67ypjdye	178	32	substantially	substantially	RB
work_55utqx7tjrft5ojtbr67ypjdye	178	33	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	178	34	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	178	35	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	178	36	three	three	CD
work_55utqx7tjrft5ojtbr67ypjdye	178	37	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	178	38	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	179	1	Character	character	NN
work_55utqx7tjrft5ojtbr67ypjdye	179	2	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	179	3	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	179	4	ratios	ratio	NNS
work_55utqx7tjrft5ojtbr67ypjdye	179	5	vary	vary	VBP
work_55utqx7tjrft5ojtbr67ypjdye	179	6	less	less	RBR
work_55utqx7tjrft5ojtbr67ypjdye	179	7	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	179	8	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	179	9	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	179	10	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	179	11	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	179	12	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	179	13	ratios	ratio	NNS
work_55utqx7tjrft5ojtbr67ypjdye	179	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	180	1	Five	five	CD
work_55utqx7tjrft5ojtbr67ypjdye	180	2	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	180	3	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	180	4	produces	produce	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	180	5	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	180	6	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	180	7	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	180	8	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	180	9	length	length	NN
work_55utqx7tjrft5ojtbr67ypjdye	180	10	near	near	IN
work_55utqx7tjrft5ojtbr67ypjdye	180	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	180	12	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	180	13	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	180	14	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	180	15	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	180	16	Lovins	Lovins	NNP
work_55utqx7tjrft5ojtbr67ypjdye	180	17	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	180	18	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	181	1	Not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	181	2	surprisingly	surprisingly	RB
work_55utqx7tjrft5ojtbr67ypjdye	181	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	181	4	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	181	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	181	6	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	181	7	produces	produce	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	181	8	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	181	9	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	181	10	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	181	11	length	length	NN
work_55utqx7tjrft5ojtbr67ypjdye	181	12	slightly	slightly	RB
work_55utqx7tjrft5ojtbr67ypjdye	181	13	less	less	JJR
work_55utqx7tjrft5ojtbr67ypjdye	181	14	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	181	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	181	16	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	181	17	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	181	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	181	19	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	181	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	181	21	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	181	22	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	181	23	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	181	24	WordNet	WordNet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	181	25	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	181	26	vary	vary	VBP
work_55utqx7tjrft5ojtbr67ypjdye	181	27	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	181	28	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	181	29	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	181	30	cor-	cor-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	181	31	pora	pora	NNP
work_55utqx7tjrft5ojtbr67ypjdye	181	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	182	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	182	2	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	182	3	verify	verify	VBP
work_55utqx7tjrft5ojtbr67ypjdye	182	4	some	some	DT
work_55utqx7tjrft5ojtbr67ypjdye	182	5	expected	expect	VBN
work_55utqx7tjrft5ojtbr67ypjdye	182	6	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	182	7	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	182	8	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	182	9	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	182	10	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	182	11	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	182	12	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	182	13	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	182	14	very	very	RB
work_55utqx7tjrft5ojtbr67ypjdye	182	15	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	182	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	182	17	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	182	18	four	four	CD
work_55utqx7tjrft5ojtbr67ypjdye	182	19	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	182	20	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	182	21	reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	182	22	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	182	23	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	182	24	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	182	25	one-	one-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	182	26	fourth	fourth	JJ
work_55utqx7tjrft5ojtbr67ypjdye	182	27	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	182	28	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	182	29	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	182	30	fifth	fifth	NN
work_55utqx7tjrft5ojtbr67ypjdye	182	31	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	182	32	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	182	33	original	original	NN
work_55utqx7tjrft5ojtbr67ypjdye	182	34	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	183	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	183	2	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	183	3	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	4	mers	mer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	183	5	behave	behave	VBP
work_55utqx7tjrft5ojtbr67ypjdye	183	6	similarly	similarly	RB
work_55utqx7tjrft5ojtbr67ypjdye	183	7	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	183	8	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	183	9	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	183	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	183	11	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	183	12	slightly	slightly	RB
work_55utqx7tjrft5ojtbr67ypjdye	183	13	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	183	14	liberal	liberal	JJ
work_55utqx7tjrft5ojtbr67ypjdye	183	15	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	16	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	183	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	183	18	Porter2	Porter2	NNP
work_55utqx7tjrft5ojtbr67ypjdye	183	19	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	20	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	183	21	Figure	figure	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	22	1	1	CD
work_55utqx7tjrft5ojtbr67ypjdye	183	23	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	183	24	Type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	25	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	183	26	token	token	VBN
work_55utqx7tjrft5ojtbr67ypjdye	183	27	ratio	ratio	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	28	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	183	29	character	character	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	30	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	183	31	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	183	32	ratio	ratio	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	33	vary	vary	VBP
work_55utqx7tjrft5ojtbr67ypjdye	183	34	substantially	substantially	RB
work_55utqx7tjrft5ojtbr67ypjdye	183	35	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	183	36	training	train	VBG
work_55utqx7tjrft5ojtbr67ypjdye	183	37	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	38	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	183	39	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	40	treat-	treat-	NN
work_55utqx7tjrft5ojtbr67ypjdye	183	41	ments	ment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	183	42	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	184	1	Due	due	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	2	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	3	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	184	4	context	context	NN
work_55utqx7tjrft5ojtbr67ypjdye	184	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	184	6	sensitive	sensitive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	184	7	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	184	8	done	do	VBN
work_55utqx7tjrft5ojtbr67ypjdye	184	9	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	184	11	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	184	12	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	184	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	184	14	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	184	15	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	184	16	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	184	17	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	18	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	184	19	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	184	20	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	184	21	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	184	22	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	184	23	map	map	VB
work_55utqx7tjrft5ojtbr67ypjdye	184	24	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	25	multiple	multiple	JJ
work_55utqx7tjrft5ojtbr67ypjdye	184	26	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	184	27	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	184	28	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	184	29	producing	produce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	184	30	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	184	31	greater	great	JJR
work_55utqx7tjrft5ojtbr67ypjdye	184	32	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	184	33	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	184	34	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	35	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	184	36	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	184	37	ratio	ratio	NN
work_55utqx7tjrft5ojtbr67ypjdye	184	38	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	39	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	184	40	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	184	41	version	version	NN
work_55utqx7tjrft5ojtbr67ypjdye	184	42	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	43	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	184	44	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	184	45	stemmer	stemmer	JJ
work_55utqx7tjrft5ojtbr67ypjdye	184	46	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	47	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	184	48	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	184	49	original	original	JJ
work_55utqx7tjrft5ojtbr67ypjdye	184	50	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	184	51	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	184	52	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	185	1	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	185	2	corpora	corpora	NNP
work_55utqx7tjrft5ojtbr67ypjdye	185	3	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	185	4	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	185	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	186	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	186	2	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	186	3	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	186	4	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	186	5	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	186	6	Lovins	Lovins	NNP
work_55utqx7tjrft5ojtbr67ypjdye	186	7	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	186	8	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	186	9	both	both	DT
work_55utqx7tjrft5ojtbr67ypjdye	186	10	stronger	strong	JJR
work_55utqx7tjrft5ojtbr67ypjdye	186	11	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	186	12	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	186	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	186	14	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	186	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	186	16	S-	s-	NN
work_55utqx7tjrft5ojtbr67ypjdye	186	17	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	186	18	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	186	19	consistently	consistently	RB
work_55utqx7tjrft5ojtbr67ypjdye	186	20	weaker	weak	JJR
work_55utqx7tjrft5ojtbr67ypjdye	186	21	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	187	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	187	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	187	3	vocabu-	vocabu-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	187	4	lary	lary	NN
work_55utqx7tjrft5ojtbr67ypjdye	187	5	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	187	6	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	187	7	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	187	8	affects	affect	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	187	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	187	10	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	187	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	187	12	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	187	13	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	187	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	187	15	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	187	16	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	187	17	little	little	JJ
work_55utqx7tjrft5ojtbr67ypjdye	187	18	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	187	19	affect	affect	VB
work_55utqx7tjrft5ojtbr67ypjdye	187	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	187	21	strengths	strength	NNS
work_55utqx7tjrft5ojtbr67ypjdye	187	22	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	187	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	187	24	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	187	25	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	187	26	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	187	27	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	187	28	relative	relative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	187	29	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	187	30	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	187	31	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	187	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	188	1	5.2	5.2	CD
work_55utqx7tjrft5ojtbr67ypjdye	188	2	Held	hold	VBN
work_55utqx7tjrft5ojtbr67ypjdye	188	3	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	188	4	Out	out	RP
work_55utqx7tjrft5ojtbr67ypjdye	188	5	Likelihood	Likelihood	NNP
work_55utqx7tjrft5ojtbr67ypjdye	188	6	Using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	188	7	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	188	8	normalized	normalized	JJ
work_55utqx7tjrft5ojtbr67ypjdye	188	9	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	188	10	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	188	11	measure	measure	NN
work_55utqx7tjrft5ojtbr67ypjdye	188	12	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	188	13	Equation	Equation	NNP
work_55utqx7tjrft5ojtbr67ypjdye	188	14	3	3	CD
work_55utqx7tjrft5ojtbr67ypjdye	188	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	188	16	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	188	17	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	188	18	compare	compare	VB
work_55utqx7tjrft5ojtbr67ypjdye	188	19	likelihoods	likelihood	NNS
work_55utqx7tjrft5ojtbr67ypjdye	188	20	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	188	21	all	all	PDT
work_55utqx7tjrft5ojtbr67ypjdye	188	22	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	188	23	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	188	24	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	188	25	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	188	26	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	188	27	shown	show	VBN
work_55utqx7tjrft5ojtbr67ypjdye	188	28	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	188	29	Figure	Figure	NNP
work_55utqx7tjrft5ojtbr67ypjdye	188	30	2	2	CD
work_55utqx7tjrft5ojtbr67ypjdye	188	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	189	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	189	2	observe	observe	VBP
work_55utqx7tjrft5ojtbr67ypjdye	189	3	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	189	4	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	189	5	standard	standard	JJ
work_55utqx7tjrft5ojtbr67ypjdye	189	6	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	7	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	189	8	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	189	9	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	189	10	treat-	treat-	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	11	ments	ment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	189	12	provide	provide	VBP
work_55utqx7tjrft5ojtbr67ypjdye	189	13	little	little	JJ
work_55utqx7tjrft5ojtbr67ypjdye	189	14	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	15	benefit	benefit	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	16	apart	apart	RB
work_55utqx7tjrft5ojtbr67ypjdye	189	17	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	189	18	reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	189	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	189	20	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	21	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	22	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	189	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	189	24	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	189	25	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	189	26	293	293	CD
work_55utqx7tjrft5ojtbr67ypjdye	189	27	Figure	figure	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	28	2	2	CD
work_55utqx7tjrft5ojtbr67ypjdye	189	29	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	189	30	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	189	31	light	light	JJ
work_55utqx7tjrft5ojtbr67ypjdye	189	32	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	33	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	189	34	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	189	35	help	help	VB
work_55utqx7tjrft5ojtbr67ypjdye	189	36	particular	particular	JJ
work_55utqx7tjrft5ojtbr67ypjdye	189	37	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	38	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	189	39	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	40	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	41	generally	generally	RB
work_55utqx7tjrft5ojtbr67ypjdye	189	42	decreases	decrease	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	189	43	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	189	44	statistical	statistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	189	45	fit	fit	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	46	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	189	47	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	189	48	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	189	49	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	50	proportionally	proportionally	RB
work_55utqx7tjrft5ojtbr67ypjdye	189	51	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	189	52	its	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	189	53	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	54	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	189	55	measured	measure	VBN
work_55utqx7tjrft5ojtbr67ypjdye	189	56	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	189	57	normalized	normalized	JJ
work_55utqx7tjrft5ojtbr67ypjdye	189	58	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	59	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	189	60	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	190	1	Con-	Con-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	190	2	fidence	fidence	NN
work_55utqx7tjrft5ojtbr67ypjdye	190	3	intervals	interval	NNS
work_55utqx7tjrft5ojtbr67ypjdye	190	4	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	190	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	190	6	p	p	NN
work_55utqx7tjrft5ojtbr67ypjdye	190	7	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	190	8	0.99	0.99	CD
work_55utqx7tjrft5ojtbr67ypjdye	190	9	range	range	NN
work_55utqx7tjrft5ojtbr67ypjdye	190	10	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	190	11	belonging	belong	VBG
work_55utqx7tjrft5ojtbr67ypjdye	190	12	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	190	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	190	14	distribution	distribution	NN
work_55utqx7tjrft5ojtbr67ypjdye	190	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	190	16	that	that	DT
work_55utqx7tjrft5ojtbr67ypjdye	190	17	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	190	18	’s	’s	POS
work_55utqx7tjrft5ojtbr67ypjdye	190	19	normalized	normalized	JJ
work_55utqx7tjrft5ojtbr67ypjdye	190	20	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	190	21	likeli-	likeli-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	190	22	hoods	hood	NNS
work_55utqx7tjrft5ojtbr67ypjdye	190	23	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	190	24	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	190	25	least	least	RBS
work_55utqx7tjrft5ojtbr67ypjdye	190	26	9	9	CD
work_55utqx7tjrft5ojtbr67ypjdye	190	27	samples	sample	NNS
work_55utqx7tjrft5ojtbr67ypjdye	190	28	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	190	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	191	1	Higher	high	JJR
work_55utqx7tjrft5ojtbr67ypjdye	191	2	values	value	NNS
work_55utqx7tjrft5ojtbr67ypjdye	191	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	191	4	normalized	normalized	JJ
work_55utqx7tjrft5ojtbr67ypjdye	191	5	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	191	6	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	191	7	represent	represent	VBP
work_55utqx7tjrft5ojtbr67ypjdye	191	8	better	well	JJR
work_55utqx7tjrft5ojtbr67ypjdye	191	9	model	model	NNP
work_55utqx7tjrft5ojtbr67ypjdye	191	10	fit	fit	NN
work_55utqx7tjrft5ojtbr67ypjdye	191	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	192	1	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	192	2	result	result	VBP
work_55utqx7tjrft5ojtbr67ypjdye	192	3	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	192	4	normalized	normalized	JJ
work_55utqx7tjrft5ojtbr67ypjdye	192	5	log	log	NN
work_55utqx7tjrft5ojtbr67ypjdye	192	6	likelihoods	likelihood	NNS
work_55utqx7tjrft5ojtbr67ypjdye	192	7	sig-	sig-	RB
work_55utqx7tjrft5ojtbr67ypjdye	192	8	nificantly	nificantly	RB
work_55utqx7tjrft5ojtbr67ypjdye	192	9	lower	low	JJR
work_55utqx7tjrft5ojtbr67ypjdye	192	10	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	192	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	192	12	unstemmed	unstemmed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	192	13	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	192	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	193	1	Sta-	sta-	UH
work_55utqx7tjrft5ojtbr67ypjdye	193	2	tistically	tistically	RB
work_55utqx7tjrft5ojtbr67ypjdye	193	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	193	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	193	5	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	193	6	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	193	7	do	do	VBP
work_55utqx7tjrft5ojtbr67ypjdye	193	8	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	193	9	appear	appear	VB
work_55utqx7tjrft5ojtbr67ypjdye	193	10	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	193	11	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	193	12	improving	improve	VBG
work_55utqx7tjrft5ojtbr67ypjdye	193	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	193	14	quality	quality	NN
work_55utqx7tjrft5ojtbr67ypjdye	193	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	193	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	193	17	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	193	18	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	193	19	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	193	20	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	193	21	merely	merely	RB
work_55utqx7tjrft5ojtbr67ypjdye	193	22	reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	193	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	193	24	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	193	25	unigrams	unigram	NNS
work_55utqx7tjrft5ojtbr67ypjdye	193	26	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	193	27	could	could	MD
work_55utqx7tjrft5ojtbr67ypjdye	193	28	generate	generate	VB
work_55utqx7tjrft5ojtbr67ypjdye	193	29	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	193	30	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	193	31	moderately	moderately	RB
work_55utqx7tjrft5ojtbr67ypjdye	193	32	principled	principle	VBN
work_55utqx7tjrft5ojtbr67ypjdye	193	33	way	way	NN
work_55utqx7tjrft5ojtbr67ypjdye	193	34	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	194	1	Both	both	DT
work_55utqx7tjrft5ojtbr67ypjdye	194	2	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	194	3	/	/	SYM
work_55utqx7tjrft5ojtbr67ypjdye	194	4	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	194	5	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	194	6	Lovins	Lovins	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	194	7	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	194	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	194	9	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	194	10	problem	problem	NN
work_55utqx7tjrft5ojtbr67ypjdye	194	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	194	12	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	194	13	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	194	14	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	194	15	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	194	16	stronger	strong	JJR
work_55utqx7tjrft5ojtbr67ypjdye	194	17	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	194	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	194	19	problems	problem	NNS
work_55utqx7tjrft5ojtbr67ypjdye	194	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	194	21	overconflation	overconflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	194	22	seem	seem	VBP
work_55utqx7tjrft5ojtbr67ypjdye	194	23	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	194	24	reduce	reduce	VB
work_55utqx7tjrft5ojtbr67ypjdye	194	25	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	194	26	quality	quality	NN
work_55utqx7tjrft5ojtbr67ypjdye	194	27	further	further	RB
work_55utqx7tjrft5ojtbr67ypjdye	194	28	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	195	1	More	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	195	2	surprising	surprising	JJ
work_55utqx7tjrft5ojtbr67ypjdye	195	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	195	4	however	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	195	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	195	6	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	195	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	195	8	mediocre	mediocre	JJ
work_55utqx7tjrft5ojtbr67ypjdye	195	9	per-	per-	NN
work_55utqx7tjrft5ojtbr67ypjdye	195	10	formance	formance	NN
work_55utqx7tjrft5ojtbr67ypjdye	195	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	195	12	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	195	13	WordNet	WordNet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	195	14	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	195	15	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	196	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	196	2	fact	fact	NN
work_55utqx7tjrft5ojtbr67ypjdye	196	3	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	196	4	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	196	5	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	196	6	IMDb	IMDb	NNS
work_55utqx7tjrft5ojtbr67ypjdye	196	7	do	do	VBP
work_55utqx7tjrft5ojtbr67ypjdye	196	8	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	196	9	see	see	VB
work_55utqx7tjrft5ojtbr67ypjdye	196	10	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	196	11	improvement	improvement	NN
work_55utqx7tjrft5ojtbr67ypjdye	196	12	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	196	13	use	use	NN
work_55utqx7tjrft5ojtbr67ypjdye	196	14	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	196	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	196	16	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	196	17	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	196	18	easy	easy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	196	19	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	196	20	explain	explain	VB
work_55utqx7tjrft5ojtbr67ypjdye	196	21	away	away	RB
work_55utqx7tjrft5ojtbr67ypjdye	196	22	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	196	23	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	196	24	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	196	25	contain	contain	NNP
work_55utqx7tjrft5ojtbr67ypjdye	196	26	slang	slang	NNP
work_55utqx7tjrft5ojtbr67ypjdye	196	27	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	196	28	misspellings	misspellings	NNP
work_55utqx7tjrft5ojtbr67ypjdye	196	29	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	196	30	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	196	31	plenty	plenty	NN
work_55utqx7tjrft5ojtbr67ypjdye	196	32	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	196	33	proper	proper	JJ
work_55utqx7tjrft5ojtbr67ypjdye	196	34	names	name	NNS
work_55utqx7tjrft5ojtbr67ypjdye	196	35	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	196	36	enough	enough	RB
work_55utqx7tjrft5ojtbr67ypjdye	196	37	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	196	38	make	make	VB
work_55utqx7tjrft5ojtbr67ypjdye	196	39	lemmatization	lemmatization	NN
work_55utqx7tjrft5ojtbr67ypjdye	196	40	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	196	41	challenge	challenge	NN
work_55utqx7tjrft5ojtbr67ypjdye	196	42	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	197	1	However	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	197	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	197	3	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	197	4	see	see	VBP
work_55utqx7tjrft5ojtbr67ypjdye	197	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	197	6	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	197	7	result	result	NN
work_55utqx7tjrft5ojtbr67ypjdye	197	8	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	197	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	197	10	case	case	NN
work_55utqx7tjrft5ojtbr67ypjdye	197	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	197	12	New	New	NNP
work_55utqx7tjrft5ojtbr67ypjdye	197	13	York	York	NNP
work_55utqx7tjrft5ojtbr67ypjdye	197	14	Times	Times	NNP
work_55utqx7tjrft5ojtbr67ypjdye	197	15	articles	article	NNS
work_55utqx7tjrft5ojtbr67ypjdye	197	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	197	17	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	197	18	ideal	ideal	JJ
work_55utqx7tjrft5ojtbr67ypjdye	197	19	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	197	20	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	197	21	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	197	22	modeling	modeling	NN
work_55utqx7tjrft5ojtbr67ypjdye	197	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	198	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	198	2	there	there	EX
work_55utqx7tjrft5ojtbr67ypjdye	198	3	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	198	4	still	still	RB
work_55utqx7tjrft5ojtbr67ypjdye	198	5	many	many	JJ
work_55utqx7tjrft5ojtbr67ypjdye	198	6	named	name	VBN
work_55utqx7tjrft5ojtbr67ypjdye	198	7	entities	entity	NNS
work_55utqx7tjrft5ojtbr67ypjdye	198	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	198	9	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	198	10	arise	arise	VBP
work_55utqx7tjrft5ojtbr67ypjdye	198	11	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	198	12	carefully	carefully	RB
work_55utqx7tjrft5ojtbr67ypjdye	198	13	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	198	14	edited	edit	VBN
work_55utqx7tjrft5ojtbr67ypjdye	198	15	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	198	16	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	198	17	stan-	stan-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	198	18	dardized	dardize	VBD
work_55utqx7tjrft5ojtbr67ypjdye	198	19	journalistic	journalistic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	198	20	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	198	21	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	199	1	Other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	199	2	observations	observation	NNS
work_55utqx7tjrft5ojtbr67ypjdye	199	3	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	199	4	less	less	RBR
work_55utqx7tjrft5ojtbr67ypjdye	199	5	surprising	surprising	JJ
work_55utqx7tjrft5ojtbr67ypjdye	199	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	200	1	Five-	five-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	200	2	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	200	3	produces	produce	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	200	4	likelihoods	likelihood	NNS
work_55utqx7tjrft5ojtbr67ypjdye	200	5	comparable	comparable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	200	6	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	200	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	200	8	stronger	strong	JJR
work_55utqx7tjrft5ojtbr67ypjdye	200	9	Lovins	Lovins	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	200	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	200	11	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	200	12	/	/	SYM
work_55utqx7tjrft5ojtbr67ypjdye	200	13	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	200	14	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	200	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	200	16	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	200	17	sig-	sig-	NN
work_55utqx7tjrft5ojtbr67ypjdye	200	18	nificantly	nificantly	RB
work_55utqx7tjrft5ojtbr67ypjdye	200	19	better	well	RBR
work_55utqx7tjrft5ojtbr67ypjdye	200	20	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	200	21	either	either	CC
work_55utqx7tjrft5ojtbr67ypjdye	200	22	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	200	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	200	24	50-topic	50-topic	CD
work_55utqx7tjrft5ojtbr67ypjdye	200	25	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	200	26	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	200	27	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	201	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	201	2	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	201	3	relate	relate	VB
work_55utqx7tjrft5ojtbr67ypjdye	201	4	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	201	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	201	6	irregularities	irregularity	NNS
work_55utqx7tjrft5ojtbr67ypjdye	201	7	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	201	8	re-	re-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	201	9	view	view	NN
work_55utqx7tjrft5ojtbr67ypjdye	201	10	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	201	11	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	201	12	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	201	13	elongated	elongate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	201	14	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	201	15	emphasis	emphasis	NN
work_55utqx7tjrft5ojtbr67ypjdye	201	16	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	201	17	e.g.	e.g.	RB
work_55utqx7tjrft5ojtbr67ypjdye	202	1	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	202	2	hel-	hel-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	202	3	loooo	loooo	NNP
work_55utqx7tjrft5ojtbr67ypjdye	202	4	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	202	5	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	202	6	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	202	7	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	202	8	oddities	oddity	NNS
work_55utqx7tjrft5ojtbr67ypjdye	202	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	202	10	online	online	JJ
work_55utqx7tjrft5ojtbr67ypjdye	202	11	informal	informal	JJ
work_55utqx7tjrft5ojtbr67ypjdye	202	12	En-	En-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	202	13	glish	glish	NNP
work_55utqx7tjrft5ojtbr67ypjdye	202	14	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	202	15	hard	hard	JJ
work_55utqx7tjrft5ojtbr67ypjdye	202	16	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	202	17	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	202	18	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	202	19	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	202	20	suffix	suffix	NN
work_55utqx7tjrft5ojtbr67ypjdye	202	21	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	202	22	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	202	23	han-	han-	RB
work_55utqx7tjrft5ojtbr67ypjdye	202	24	dle	dle	VB
work_55utqx7tjrft5ojtbr67ypjdye	202	25	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	202	26	still	still	RB
work_55utqx7tjrft5ojtbr67ypjdye	202	27	benefit	benefit	VBP
work_55utqx7tjrft5ojtbr67ypjdye	202	28	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	202	29	naı̈ve	naı̈ve	NNP
work_55utqx7tjrft5ojtbr67ypjdye	202	30	forms	form	NNS
work_55utqx7tjrft5ojtbr67ypjdye	202	31	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	202	32	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	202	33	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	203	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	203	2	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	203	3	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	203	4	Porter2	Porter2	NNP
work_55utqx7tjrft5ojtbr67ypjdye	203	5	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	203	6	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	203	7	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	203	8	signifi-	signifi-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	203	9	cantly	cantly	RB
work_55utqx7tjrft5ojtbr67ypjdye	203	10	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	203	11	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	203	12	any	any	DT
work_55utqx7tjrft5ojtbr67ypjdye	203	13	case	case	NN
work_55utqx7tjrft5ojtbr67ypjdye	203	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	203	15	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	203	16	serves	serve	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	203	17	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	203	18	com-	com-	NN
work_55utqx7tjrft5ojtbr67ypjdye	203	19	forting	fort	VBG
work_55utqx7tjrft5ojtbr67ypjdye	203	20	validation	validation	NN
work_55utqx7tjrft5ojtbr67ypjdye	203	21	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	203	22	those	those	DT
work_55utqx7tjrft5ojtbr67ypjdye	203	23	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	203	24	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	203	25	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	203	26	new	new	JJ
work_55utqx7tjrft5ojtbr67ypjdye	203	27	gen-	gen-	NN
work_55utqx7tjrft5ojtbr67ypjdye	203	28	eration	eration	NN
work_55utqx7tjrft5ojtbr67ypjdye	203	29	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	203	30	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	203	31	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	203	32	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	203	33	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	203	34	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	203	35	losing	lose	VBG
work_55utqx7tjrft5ojtbr67ypjdye	203	36	much	much	RB
work_55utqx7tjrft5ojtbr67ypjdye	203	37	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	204	1	5.3	5.3	CD
work_55utqx7tjrft5ojtbr67ypjdye	204	2	Topic	Topic	NNP
work_55utqx7tjrft5ojtbr67ypjdye	204	3	Coherence	Coherence	NNP
work_55utqx7tjrft5ojtbr67ypjdye	204	4	Log	Log	NNP
work_55utqx7tjrft5ojtbr67ypjdye	204	5	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	204	6	measures	measure	NNS
work_55utqx7tjrft5ojtbr67ypjdye	204	7	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	204	8	tell	tell	VB
work_55utqx7tjrft5ojtbr67ypjdye	204	9	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	204	10	about	about	IN
work_55utqx7tjrft5ojtbr67ypjdye	204	11	statisti-	statisti-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	204	12	cal	cal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	204	13	fit	fit	NNP
work_55utqx7tjrft5ojtbr67ypjdye	204	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	204	15	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	204	16	do	do	VBP
work_55utqx7tjrft5ojtbr67ypjdye	204	17	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	204	18	necessarily	necessarily	RB
work_55utqx7tjrft5ojtbr67ypjdye	204	19	tell	tell	VB
work_55utqx7tjrft5ojtbr67ypjdye	204	20	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	204	21	about	about	IN
work_55utqx7tjrft5ojtbr67ypjdye	204	22	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	204	23	actual	actual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	204	24	apparent	apparent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	204	25	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	204	26	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	204	27	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	204	28	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	204	29	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	204	30	terms	term	NNS
work_55utqx7tjrft5ojtbr67ypjdye	204	31	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	204	32	concep-	concep-	NN
work_55utqx7tjrft5ojtbr67ypjdye	204	33	tual	tual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	204	34	similarity	similarity	NN
work_55utqx7tjrft5ojtbr67ypjdye	204	35	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	204	36	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	204	37	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	204	38	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	204	39	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	204	40	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	205	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	205	2	Figure	figure	NN
work_55utqx7tjrft5ojtbr67ypjdye	205	3	3	3	CD
work_55utqx7tjrft5ojtbr67ypjdye	205	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	205	5	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	205	6	display	display	VBP
work_55utqx7tjrft5ojtbr67ypjdye	205	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	205	8	negative	negative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	205	9	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	205	10	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	205	11	scores	score	NNS
work_55utqx7tjrft5ojtbr67ypjdye	205	12	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	205	13	Equation	Equation	NNP
work_55utqx7tjrft5ojtbr67ypjdye	205	14	4	4	CD
work_55utqx7tjrft5ojtbr67ypjdye	205	15	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	205	16	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	205	17	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	205	18	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	206	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	206	2	hypothesis	hypothesis	NN
work_55utqx7tjrft5ojtbr67ypjdye	206	3	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	206	4	test	test	VBP
work_55utqx7tjrft5ojtbr67ypjdye	206	5	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	206	6	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	206	7	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	206	8	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	206	9	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	206	10	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	206	11	should	should	MD
work_55utqx7tjrft5ojtbr67ypjdye	206	12	map	map	VB
work_55utqx7tjrft5ojtbr67ypjdye	206	13	morphologically	morphologically	RB
work_55utqx7tjrft5ojtbr67ypjdye	206	14	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	206	15	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	206	16	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	206	17	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	206	18	shared	shared	JJ
work_55utqx7tjrft5ojtbr67ypjdye	206	19	con-	con-	NN
work_55utqx7tjrft5ojtbr67ypjdye	206	20	cept	cept	VBD
work_55utqx7tjrft5ojtbr67ypjdye	206	21	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	206	22	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	206	23	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	206	24	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	206	25	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	206	26	automatically	automatically	RB
work_55utqx7tjrft5ojtbr67ypjdye	206	27	constraining	constrain	VBG
work_55utqx7tjrft5ojtbr67ypjdye	206	28	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	206	29	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	206	30	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	206	31	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	206	32	ensure	ensure	VB
work_55utqx7tjrft5ojtbr67ypjdye	206	33	closely	closely	RB
work_55utqx7tjrft5ojtbr67ypjdye	206	34	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	206	35	related	relate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	206	36	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	206	37	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	206	38	proportionally	proportionally	RB
work_55utqx7tjrft5ojtbr67ypjdye	206	39	present	present	JJ
work_55utqx7tjrft5ojtbr67ypjdye	206	40	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	206	41	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	206	42	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	206	43	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	206	44	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	207	1	Our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	207	2	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	207	3	do	do	VBP
work_55utqx7tjrft5ojtbr67ypjdye	207	4	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	207	5	conform	conform	VB
work_55utqx7tjrft5ojtbr67ypjdye	207	6	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	207	7	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	207	8	intuition	intuition	NN
work_55utqx7tjrft5ojtbr67ypjdye	207	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	208	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	208	2	majority	majority	NN
work_55utqx7tjrft5ojtbr67ypjdye	208	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	208	4	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	208	5	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	208	6	statistically	statistically	RB
work_55utqx7tjrft5ojtbr67ypjdye	208	7	indistin-	indistin-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	208	8	guishable	guishable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	208	9	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	208	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	208	11	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	208	12	control	control	NN
work_55utqx7tjrft5ojtbr67ypjdye	208	13	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	208	14	respect	respect	NN
work_55utqx7tjrft5ojtbr67ypjdye	208	15	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	208	16	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	208	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	209	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	209	2	relative	relative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	209	3	effects	effect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	209	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	209	5	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	209	6	treat-	treat-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	209	7	ments	ment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	209	8	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	209	9	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	209	10	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	209	11	magnified	magnify	VBN
work_55utqx7tjrft5ojtbr67ypjdye	209	12	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	209	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	209	14	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	209	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	209	16	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	209	17	increases	increase	NNS
work_55utqx7tjrft5ojtbr67ypjdye	209	18	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	209	19	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	209	20	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	209	21	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	209	22	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	209	23	dif-	dif-	CC
work_55utqx7tjrft5ojtbr67ypjdye	209	24	fers	fer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	209	25	significantly	significantly	RB
work_55utqx7tjrft5ojtbr67ypjdye	209	26	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	209	27	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	209	28	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	209	29	10	10	CD
work_55utqx7tjrft5ojtbr67ypjdye	209	30	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	209	31	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	209	32	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	209	33	200	200	CD
work_55utqx7tjrft5ojtbr67ypjdye	209	34	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	209	35	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	209	36	four	four	CD
work_55utqx7tjrft5ojtbr67ypjdye	209	37	strongest	strong	JJS
work_55utqx7tjrft5ojtbr67ypjdye	209	38	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	209	39	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	209	40	Lovins	Lovins	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	209	41	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	209	42	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	209	43	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	209	44	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	209	45	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	209	46	five	five	CD
work_55utqx7tjrft5ojtbr67ypjdye	209	47	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	209	48	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	209	49	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	209	50	four	four	CD
work_55utqx7tjrft5ojtbr67ypjdye	209	51	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	209	52	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	209	53	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	209	54	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	209	55	significantly	significantly	RB
work_55utqx7tjrft5ojtbr67ypjdye	209	56	worse	bad	JJR
work_55utqx7tjrft5ojtbr67ypjdye	209	57	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	210	1	Four	four	CD
work_55utqx7tjrft5ojtbr67ypjdye	210	2	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	210	3	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	210	4	suffers	suffer	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	210	5	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	210	6	similar	similar	JJ
work_55utqx7tjrft5ojtbr67ypjdye	210	7	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	210	8	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	210	9	IMDb	IMDb	NNS
work_55utqx7tjrft5ojtbr67ypjdye	210	10	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	210	11	50	50	CD
work_55utqx7tjrft5ojtbr67ypjdye	210	12	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	210	13	200	200	CD
work_55utqx7tjrft5ojtbr67ypjdye	210	14	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	210	15	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	211	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	211	2	contrast	contrast	NN
work_55utqx7tjrft5ojtbr67ypjdye	211	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	211	4	four-	four-	NN
work_55utqx7tjrft5ojtbr67ypjdye	211	5	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	211	6	actually	actually	RB
work_55utqx7tjrft5ojtbr67ypjdye	211	7	improves	improve	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	211	8	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	211	9	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	211	10	compared	compare	VBN
work_55utqx7tjrft5ojtbr67ypjdye	211	11	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	211	12	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	211	13	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	211	14	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	211	15	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	211	16	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	211	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	211	18	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	211	19	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	211	20	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	211	21	increases	increase	NNS
work_55utqx7tjrft5ojtbr67ypjdye	211	22	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	211	23	reaching	reach	VBG
work_55utqx7tjrft5ojtbr67ypjdye	211	24	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	211	25	significant	significant	JJ
work_55utqx7tjrft5ojtbr67ypjdye	211	26	level	level	NN
work_55utqx7tjrft5ojtbr67ypjdye	211	27	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	211	28	200	200	CD
work_55utqx7tjrft5ojtbr67ypjdye	211	29	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	211	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	212	1	294	294	CD
work_55utqx7tjrft5ojtbr67ypjdye	212	2	Figure	figure	NN
work_55utqx7tjrft5ojtbr67ypjdye	212	3	3	3	CD
work_55utqx7tjrft5ojtbr67ypjdye	212	4	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	212	5	Conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	212	6	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	212	7	introduce	introduce	VBP
work_55utqx7tjrft5ojtbr67ypjdye	212	8	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	212	9	significant	significant	JJ
work_55utqx7tjrft5ojtbr67ypjdye	212	10	difference	difference	NN
work_55utqx7tjrft5ojtbr67ypjdye	212	11	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	212	12	almost	almost	RB
work_55utqx7tjrft5ojtbr67ypjdye	212	13	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	212	14	cases	case	NNS
work_55utqx7tjrft5ojtbr67ypjdye	212	15	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	212	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	212	17	resulting	result	VBG
work_55utqx7tjrft5ojtbr67ypjdye	212	18	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	212	19	neg-	neg-	NN
work_55utqx7tjrft5ojtbr67ypjdye	212	20	ative	ative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	212	21	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	212	22	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	212	23	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	212	24	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	212	25	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	212	26	according	accord	VBG
work_55utqx7tjrft5ojtbr67ypjdye	212	27	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	212	28	token	token	JJ
work_55utqx7tjrft5ojtbr67ypjdye	212	29	assignments	assignment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	212	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	213	1	Smaller	small	JJR
work_55utqx7tjrft5ojtbr67ypjdye	213	2	values	value	NNS
work_55utqx7tjrft5ojtbr67ypjdye	213	3	indicating	indicate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	213	4	better	well	JJR
work_55utqx7tjrft5ojtbr67ypjdye	213	5	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	213	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	213	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	213	8	error	error	NN
work_55utqx7tjrft5ojtbr67ypjdye	213	9	bars	bar	NNS
work_55utqx7tjrft5ojtbr67ypjdye	213	10	represent	represent	VBP
work_55utqx7tjrft5ojtbr67ypjdye	213	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	213	12	p	p	NN
work_55utqx7tjrft5ojtbr67ypjdye	213	13	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	213	14	0.99	0.99	CD
work_55utqx7tjrft5ojtbr67ypjdye	213	15	range	range	NN
work_55utqx7tjrft5ojtbr67ypjdye	213	16	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	213	17	possible	possible	JJ
work_55utqx7tjrft5ojtbr67ypjdye	213	18	mean	mean	JJ
work_55utqx7tjrft5ojtbr67ypjdye	213	19	values	value	NNS
work_55utqx7tjrft5ojtbr67ypjdye	213	20	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	214	1	Given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	214	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	214	3	lack	lack	NN
work_55utqx7tjrft5ojtbr67ypjdye	214	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	214	5	substantial	substantial	JJ
work_55utqx7tjrft5ojtbr67ypjdye	214	6	statistical	statistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	214	7	difference	difference	NN
work_55utqx7tjrft5ojtbr67ypjdye	214	8	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	214	9	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	214	10	variety	variety	NN
work_55utqx7tjrft5ojtbr67ypjdye	214	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	214	12	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	214	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	214	14	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	214	15	seems	seem	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	214	16	safe	safe	JJ
work_55utqx7tjrft5ojtbr67ypjdye	214	17	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	214	18	con-	con-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	214	19	clude	clude	VB
work_55utqx7tjrft5ojtbr67ypjdye	214	20	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	214	21	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	214	22	use	use	NN
work_55utqx7tjrft5ojtbr67ypjdye	214	23	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	214	24	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	214	25	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	214	26	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	214	27	substantially	substantially	RB
work_55utqx7tjrft5ojtbr67ypjdye	214	28	improving	improve	VBG
work_55utqx7tjrft5ojtbr67ypjdye	214	29	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	214	30	encoding	encoding	NN
work_55utqx7tjrft5ojtbr67ypjdye	214	31	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	214	32	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	214	33	similarities	similarity	NNS
work_55utqx7tjrft5ojtbr67ypjdye	214	34	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	214	35	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	214	36	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	214	37	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	215	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	215	2	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	215	3	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	215	4	itself	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	215	5	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	215	6	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	215	7	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	215	8	cor-	cor-	NN
work_55utqx7tjrft5ojtbr67ypjdye	215	9	pus	pus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	215	10	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	215	11	perhaps	perhaps	RB
work_55utqx7tjrft5ojtbr67ypjdye	215	12	already	already	RB
work_55utqx7tjrft5ojtbr67ypjdye	215	13	doing	do	VBG
work_55utqx7tjrft5ojtbr67ypjdye	215	14	as	as	RB
work_55utqx7tjrft5ojtbr67ypjdye	215	15	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	215	16	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	215	17	job	job	NN
work_55utqx7tjrft5ojtbr67ypjdye	215	18	ensuring	ensure	VBG
work_55utqx7tjrft5ojtbr67ypjdye	215	19	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	215	20	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	215	21	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	215	22	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	215	23	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	215	24	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	215	25	class	class	NN
work_55utqx7tjrft5ojtbr67ypjdye	215	26	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	215	27	statisti-	statisti-	NN
work_55utqx7tjrft5ojtbr67ypjdye	215	28	cally	cally	RB
work_55utqx7tjrft5ojtbr67ypjdye	215	29	similar	similar	JJ
work_55utqx7tjrft5ojtbr67ypjdye	215	30	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	215	31	distributions	distribution	NNS
work_55utqx7tjrft5ojtbr67ypjdye	215	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	216	1	Unstemmed	unstemmed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	216	2	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	216	3	sometimes	sometimes	RB
work_55utqx7tjrft5ojtbr67ypjdye	216	4	contain	contain	VBP
work_55utqx7tjrft5ojtbr67ypjdye	216	5	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	216	6	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	216	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	216	8	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	216	9	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	216	10	class	class	NN
work_55utqx7tjrft5ojtbr67ypjdye	216	11	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	216	12	e.g.	e.g.	RB
work_55utqx7tjrft5ojtbr67ypjdye	217	1	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	217	2	restaurant	restaurant	NN
work_55utqx7tjrft5ojtbr67ypjdye	217	3	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	217	4	versus	versus	IN
work_55utqx7tjrft5ojtbr67ypjdye	217	5	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	217	6	restaurants	restaurant	NNS
work_55utqx7tjrft5ojtbr67ypjdye	217	7	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	217	8	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	217	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	218	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	218	2	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	218	3	might	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	218	4	give	give	VB
work_55utqx7tjrft5ojtbr67ypjdye	218	5	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	218	6	slight	slight	JJ
work_55utqx7tjrft5ojtbr67ypjdye	218	7	ad-	ad-	NN
work_55utqx7tjrft5ojtbr67ypjdye	218	8	vantage	vantage	NN
work_55utqx7tjrft5ojtbr67ypjdye	218	9	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	218	10	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	218	11	measures	measure	NNS
work_55utqx7tjrft5ojtbr67ypjdye	218	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	218	13	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	218	14	case	case	NN
work_55utqx7tjrft5ojtbr67ypjdye	218	15	implies	imply	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	218	16	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	218	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	218	18	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	218	19	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	218	20	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	218	21	necessary	necessary	JJ
work_55utqx7tjrft5ojtbr67ypjdye	218	22	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	218	23	bringing	bring	VBG
work_55utqx7tjrft5ojtbr67ypjdye	218	24	to-	to-	XX
work_55utqx7tjrft5ojtbr67ypjdye	218	25	gether	gether	JJ
work_55utqx7tjrft5ojtbr67ypjdye	218	26	morphological	morphological	JJ
work_55utqx7tjrft5ojtbr67ypjdye	218	27	variants	variant	NNS
work_55utqx7tjrft5ojtbr67ypjdye	218	28	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	218	29	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	218	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	219	1	5.4	5.4	CD
work_55utqx7tjrft5ojtbr67ypjdye	219	2	Clustering	Clustering	NNP
work_55utqx7tjrft5ojtbr67ypjdye	219	3	Consistency	Consistency	NNP
work_55utqx7tjrft5ojtbr67ypjdye	219	4	Another	another	DT
work_55utqx7tjrft5ojtbr67ypjdye	219	5	hypothesized	hypothesize	VBN
work_55utqx7tjrft5ojtbr67ypjdye	219	6	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	219	7	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	219	8	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	219	9	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	219	10	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	219	11	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	219	12	will	will	MD
work_55utqx7tjrft5ojtbr67ypjdye	219	13	produce	produce	VB
work_55utqx7tjrft5ojtbr67ypjdye	219	14	more	more	JJR
work_55utqx7tjrft5ojtbr67ypjdye	219	15	consistent	consistent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	219	16	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	219	17	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	219	18	reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	219	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	219	20	sensitivity	sensitivity	NN
work_55utqx7tjrft5ojtbr67ypjdye	219	21	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	219	22	related	related	JJ
work_55utqx7tjrft5ojtbr67ypjdye	219	23	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	219	24	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	219	25	random	random	JJ
work_55utqx7tjrft5ojtbr67ypjdye	219	26	initialization	initialization	NN
work_55utqx7tjrft5ojtbr67ypjdye	219	27	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	220	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	220	2	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	220	3	use	use	VB
work_55utqx7tjrft5ojtbr67ypjdye	220	4	variation	variation	NN
work_55utqx7tjrft5ojtbr67ypjdye	220	5	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	220	6	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	220	7	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	220	8	VOI	VOI	NNP
work_55utqx7tjrft5ojtbr67ypjdye	220	9	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	220	10	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	220	11	un-	un-	XX
work_55utqx7tjrft5ojtbr67ypjdye	220	12	derstand	derstand	NN
work_55utqx7tjrft5ojtbr67ypjdye	220	13	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	220	14	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	220	15	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	220	16	differ	differ	VBP
work_55utqx7tjrft5ojtbr67ypjdye	220	17	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	220	18	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	220	19	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	220	20	relative	relative	NN
work_55utqx7tjrft5ojtbr67ypjdye	220	21	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	220	22	how	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	220	23	much	much	JJ
work_55utqx7tjrft5ojtbr67ypjdye	220	24	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	220	25	vary	vary	VBP
work_55utqx7tjrft5ojtbr67ypjdye	220	26	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	220	27	random	random	JJ
work_55utqx7tjrft5ojtbr67ypjdye	220	28	ini-	ini-	DT
work_55utqx7tjrft5ojtbr67ypjdye	220	29	tializations	tialization	NNS
work_55utqx7tjrft5ojtbr67ypjdye	220	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	221	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	221	2	summarize	summarize	VBP
work_55utqx7tjrft5ojtbr67ypjdye	221	3	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	221	4	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	221	5	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	221	6	Figure	figure	NN
work_55utqx7tjrft5ojtbr67ypjdye	221	7	4	4	CD
work_55utqx7tjrft5ojtbr67ypjdye	221	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	222	1	Within	within	IN
work_55utqx7tjrft5ojtbr67ypjdye	222	2	statistical	statistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	222	3	error	error	NN
work_55utqx7tjrft5ojtbr67ypjdye	222	4	bounds	bound	NNS
work_55utqx7tjrft5ojtbr67ypjdye	222	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	222	6	intra	intra	JJ
work_55utqx7tjrft5ojtbr67ypjdye	222	7	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	222	8	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	222	9	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	222	10	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	222	11	always	always	RB
work_55utqx7tjrft5ojtbr67ypjdye	222	12	less	less	JJR
work_55utqx7tjrft5ojtbr67ypjdye	222	13	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	222	14	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	222	15	equal	equal	JJ
work_55utqx7tjrft5ojtbr67ypjdye	222	16	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	222	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	222	18	variation	variation	NN
work_55utqx7tjrft5ojtbr67ypjdye	222	19	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	222	20	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	222	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	222	22	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	222	23	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	222	24	increases	increase	NNS
work_55utqx7tjrft5ojtbr67ypjdye	222	25	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	222	26	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	222	27	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	222	28	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	222	29	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	222	30	increases	increase	NNS
work_55utqx7tjrft5ojtbr67ypjdye	222	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	223	1	On	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	223	2	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	223	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	223	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	223	5	light	light	JJ
work_55utqx7tjrft5ojtbr67ypjdye	223	6	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	223	7	—	—	:
work_55utqx7tjrft5ojtbr67ypjdye	223	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	223	9	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	223	10	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	223	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	223	12	S	s	NN
work_55utqx7tjrft5ojtbr67ypjdye	223	13	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	223	14	stemmer	stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	223	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	223	16	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	223	17	WordNet	WordNet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	223	18	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	223	19	—	—	:
work_55utqx7tjrft5ojtbr67ypjdye	223	20	behave	behave	VBP
work_55utqx7tjrft5ojtbr67ypjdye	223	21	indistinguishably	indistinguishably	RB
work_55utqx7tjrft5ojtbr67ypjdye	223	22	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	223	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	223	24	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	223	25	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	223	26	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	224	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	224	2	intra	intra	JJ
work_55utqx7tjrft5ojtbr67ypjdye	224	3	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	224	4	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	224	5	VOI	VOI	NNP
work_55utqx7tjrft5ojtbr67ypjdye	224	6	trend	trend	NN
work_55utqx7tjrft5ojtbr67ypjdye	224	7	shows	show	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	224	8	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	224	9	stronger	strong	JJR
work_55utqx7tjrft5ojtbr67ypjdye	224	10	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	224	11	generally	generally	RB
work_55utqx7tjrft5ojtbr67ypjdye	224	12	result	result	VBP
work_55utqx7tjrft5ojtbr67ypjdye	224	13	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	224	14	less	less	RBR
work_55utqx7tjrft5ojtbr67ypjdye	224	15	consistent	consistent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	224	16	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	224	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	225	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	225	2	contradicts	contradict	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	225	3	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	225	4	intuition	intuition	NN
work_55utqx7tjrft5ojtbr67ypjdye	225	5	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	225	6	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	225	7	will	will	MD
work_55utqx7tjrft5ojtbr67ypjdye	225	8	help	help	VB
work_55utqx7tjrft5ojtbr67ypjdye	225	9	place	place	VB
work_55utqx7tjrft5ojtbr67ypjdye	225	10	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	225	11	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	225	12	similar	similar	JJ
work_55utqx7tjrft5ojtbr67ypjdye	225	13	meaning	meaning	NN
work_55utqx7tjrft5ojtbr67ypjdye	225	14	into	into	IN
work_55utqx7tjrft5ojtbr67ypjdye	225	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	225	16	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	225	17	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	225	18	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	226	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	226	2	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	226	3	con-	con-	NN
work_55utqx7tjrft5ojtbr67ypjdye	226	4	strains	strain	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	226	5	all	all	DT
work_55utqx7tjrft5ojtbr67ypjdye	226	6	conflated	conflate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	226	7	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	226	8	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	226	9	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	226	10	share	share	VB
work_55utqx7tjrft5ojtbr67ypjdye	226	11	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	226	12	prob-	prob-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	226	13	ability	ability	NN
work_55utqx7tjrft5ojtbr67ypjdye	226	14	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	226	15	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	226	16	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	226	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	226	18	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	226	19	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	226	20	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	226	21	ensure	ensure	VB
work_55utqx7tjrft5ojtbr67ypjdye	226	22	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	226	23	those	those	DT
work_55utqx7tjrft5ojtbr67ypjdye	226	24	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	226	25	distributions	distribution	NNS
work_55utqx7tjrft5ojtbr67ypjdye	226	26	will	will	MD
work_55utqx7tjrft5ojtbr67ypjdye	226	27	favor	favor	VB
work_55utqx7tjrft5ojtbr67ypjdye	226	28	few	few	JJ
work_55utqx7tjrft5ojtbr67ypjdye	226	29	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	226	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	227	1	There	there	EX
work_55utqx7tjrft5ojtbr67ypjdye	227	2	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	227	3	two	two	CD
work_55utqx7tjrft5ojtbr67ypjdye	227	4	striking	strike	VBG
work_55utqx7tjrft5ojtbr67ypjdye	227	5	exceptions	exception	NNS
work_55utqx7tjrft5ojtbr67ypjdye	227	6	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	227	7	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	227	8	trend	trend	NN
work_55utqx7tjrft5ojtbr67ypjdye	227	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	228	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	228	2	first	first	JJ
work_55utqx7tjrft5ojtbr67ypjdye	228	3	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	228	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	228	5	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	228	6	stemmer	stemmer	JJ
work_55utqx7tjrft5ojtbr67ypjdye	228	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	229	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	229	2	intra	intra	JJ
work_55utqx7tjrft5ojtbr67ypjdye	229	3	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	229	4	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	229	5	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	229	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	229	7	Krovetz	Krovetz	NNP
work_55utqx7tjrft5ojtbr67ypjdye	229	8	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	229	9	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	229	10	stays	stay	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	229	11	closer	close	RBR
work_55utqx7tjrft5ojtbr67ypjdye	229	12	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	229	13	that	that	DT
work_55utqx7tjrft5ojtbr67ypjdye	229	14	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	229	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	229	16	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	229	17	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	229	18	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	229	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	229	20	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	229	21	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	229	22	stemmer	stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	229	23	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	229	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	229	25	WordNet	WordNet	NNP
work_55utqx7tjrft5ojtbr67ypjdye	229	26	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	229	27	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	230	1	However	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	230	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	230	3	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	230	4	higher	high	JJR
work_55utqx7tjrft5ojtbr67ypjdye	230	5	inter-	inter-	FW
work_55utqx7tjrft5ojtbr67ypjdye	230	6	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	230	7	VOI	VOI	NNP
work_55utqx7tjrft5ojtbr67ypjdye	230	8	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	230	9	Krovetz	Krovetz	NNP
work_55utqx7tjrft5ojtbr67ypjdye	230	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	230	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	230	12	unstemmed	unstemmed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	230	13	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	230	14	suggests	suggest	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	230	15	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	230	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	230	17	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	230	18	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	230	19	produces	produce	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	230	20	small	small	JJ
work_55utqx7tjrft5ojtbr67ypjdye	230	21	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	230	22	significant	significant	JJ
work_55utqx7tjrft5ojtbr67ypjdye	230	23	changes	change	NNS
work_55utqx7tjrft5ojtbr67ypjdye	230	24	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	230	25	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	230	26	optima	optima	NN
work_55utqx7tjrft5ojtbr67ypjdye	230	27	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	230	28	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	230	29	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	230	30	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	230	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	231	1	For	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	231	2	IMDb	IMDb	NNS
work_55utqx7tjrft5ojtbr67ypjdye	231	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	231	4	NYT	NYT	NNP
work_55utqx7tjrft5ojtbr67ypjdye	231	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	231	6	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	231	7	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	231	8	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	231	9	200	200	CD
work_55utqx7tjrft5ojtbr67ypjdye	231	10	top-	top-	NN
work_55utqx7tjrft5ojtbr67ypjdye	231	11	ics	ics	NN
work_55utqx7tjrft5ojtbr67ypjdye	231	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	231	13	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	231	14	NYT	NYT	NNP
work_55utqx7tjrft5ojtbr67ypjdye	231	15	again	again	RB
work_55utqx7tjrft5ojtbr67ypjdye	231	16	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	231	17	50	50	CD
work_55utqx7tjrft5ojtbr67ypjdye	231	18	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	231	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	231	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	231	21	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	231	22	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	231	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	231	24	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	231	25	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	231	26	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	231	27	Krovetz	Krovetz	NNP
work_55utqx7tjrft5ojtbr67ypjdye	231	28	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	231	29	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	231	30	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	231	31	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	231	32	significantly	significantly	RB
work_55utqx7tjrft5ojtbr67ypjdye	231	33	greater	great	JJR
work_55utqx7tjrft5ojtbr67ypjdye	231	34	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	231	35	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	231	36	VOI	VOI	NNP
work_55utqx7tjrft5ojtbr67ypjdye	231	37	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	231	38	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	231	39	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	231	40	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	231	41	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	231	42	itself	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	231	43	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	232	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	232	2	contrast	contrast	NN
work_55utqx7tjrft5ojtbr67ypjdye	232	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	232	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	232	5	variation	variation	NN
work_55utqx7tjrft5ojtbr67ypjdye	232	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	232	7	infor-	infor-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	232	8	mation	mation	NN
work_55utqx7tjrft5ojtbr67ypjdye	232	9	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	232	10	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	232	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	232	12	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	232	13	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	232	14	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	232	15	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	232	16	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	232	17	only	only	RB
work_55utqx7tjrft5ojtbr67ypjdye	232	18	negligibly	negligibly	RB
work_55utqx7tjrft5ojtbr67ypjdye	232	19	higher	high	JJR
work_55utqx7tjrft5ojtbr67ypjdye	232	20	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	232	21	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	232	22	intra	intra	JJ
work_55utqx7tjrft5ojtbr67ypjdye	232	23	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	232	24	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	232	25	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	232	26	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	232	27	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	232	28	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	232	29	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	232	30	stemmer	stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	232	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	233	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	233	2	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	233	3	interesting	interesting	JJ
work_55utqx7tjrft5ojtbr67ypjdye	233	4	given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	233	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	233	6	repu-	repu-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	233	7	tation	tation	NN
work_55utqx7tjrft5ojtbr67ypjdye	233	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	233	9	Krovetz	Krovetz	NNP
work_55utqx7tjrft5ojtbr67ypjdye	233	10	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	233	11	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	233	12	weak	weak	JJ
work_55utqx7tjrft5ojtbr67ypjdye	233	13	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	233	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	234	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	234	2	second	second	JJ
work_55utqx7tjrft5ojtbr67ypjdye	234	3	exception	exception	NN
work_55utqx7tjrft5ojtbr67ypjdye	234	4	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	234	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	234	6	five	five	CD
work_55utqx7tjrft5ojtbr67ypjdye	234	7	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	234	8	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	234	9	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	234	10	mer	mer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	234	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	235	1	Though	though	IN
work_55utqx7tjrft5ojtbr67ypjdye	235	2	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	235	3	very	very	RB
work_55utqx7tjrft5ojtbr67ypjdye	235	4	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	235	5	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	235	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	235	7	its	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	235	8	VOI	voi	NN
work_55utqx7tjrft5ojtbr67ypjdye	235	9	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	235	10	in-	in-	RB
work_55utqx7tjrft5ojtbr67ypjdye	235	11	distinguishable	distinguishable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	235	12	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	235	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	235	14	heavier	heavy	JJR
work_55utqx7tjrft5ojtbr67ypjdye	235	15	Lovins	Lovins	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	235	16	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	235	17	Paice-	Paice-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	235	18	Husk	Husk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	235	19	stemmers	stemmer	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	235	20	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	235	21	most	most	JJS
work_55utqx7tjrft5ojtbr67ypjdye	235	22	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	235	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	235	24	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	235	25	when	when	WRB
work_55utqx7tjrft5ojtbr67ypjdye	235	26	applied	apply	VBN
work_55utqx7tjrft5ojtbr67ypjdye	235	27	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	235	28	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	235	29	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	235	30	200	200	CD
work_55utqx7tjrft5ojtbr67ypjdye	235	31	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	235	32	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	235	33	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	235	34	actually	actually	RB
work_55utqx7tjrft5ojtbr67ypjdye	235	35	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	235	36	significantly	significantly	RB
work_55utqx7tjrft5ojtbr67ypjdye	235	37	better	well	JJR
work_55utqx7tjrft5ojtbr67ypjdye	235	38	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	235	39	either	either	RB
work_55utqx7tjrft5ojtbr67ypjdye	235	40	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	235	41	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	235	42	both	both	DT
work_55utqx7tjrft5ojtbr67ypjdye	235	43	intra	intra	JJ
work_55utqx7tjrft5ojtbr67ypjdye	235	44	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	235	45	treatment	treatment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	235	46	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	235	47	inter-	inter-	NN
work_55utqx7tjrft5ojtbr67ypjdye	235	48	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	235	49	VOI	VOI	NNP
work_55utqx7tjrft5ojtbr67ypjdye	235	50	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	235	51	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	235	52	unstemmed	unstemmed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	235	53	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	235	54	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	236	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	236	2	ef-	ef-	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	3	fect	fect	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	4	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	236	5	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	236	6	seen	see	VBN
work_55utqx7tjrft5ojtbr67ypjdye	236	7	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	236	8	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	236	9	less	less	RBR
work_55utqx7tjrft5ojtbr67ypjdye	236	10	significant	significant	JJ
work_55utqx7tjrft5ojtbr67ypjdye	236	11	extent	extent	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	12	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	236	13	mod-	mod-	DT
work_55utqx7tjrft5ojtbr67ypjdye	236	14	295	295	CD
work_55utqx7tjrft5ojtbr67ypjdye	236	15	Figure	figure	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	16	4	4	CD
work_55utqx7tjrft5ojtbr67ypjdye	236	17	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	236	18	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	236	19	variation	variation	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	236	21	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	22	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	236	23	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	236	24	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	236	25	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	236	26	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	27	indicates	indicate	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	236	28	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	236	29	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	236	30	light	light	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	31	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	236	32	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	236	33	improve	improve	VB
work_55utqx7tjrft5ojtbr67ypjdye	236	34	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	236	35	comparative	comparative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	236	36	similarity	similarity	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	37	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	236	38	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	236	39	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	236	40	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	236	41	heavier	heavy	JJR
work_55utqx7tjrft5ojtbr67ypjdye	236	42	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	236	43	produce	produce	VBP
work_55utqx7tjrft5ojtbr67ypjdye	236	44	less	less	RBR
work_55utqx7tjrft5ojtbr67ypjdye	236	45	stable	stable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	236	46	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	236	47	assignments	assignment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	236	48	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	237	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	237	2	minimum	minimum	NN
work_55utqx7tjrft5ojtbr67ypjdye	237	3	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	237	4	statistical	statistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	237	5	significance	significance	NN
work_55utqx7tjrft5ojtbr67ypjdye	237	6	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	237	7	computed	compute	VBN
work_55utqx7tjrft5ojtbr67ypjdye	237	8	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	237	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	237	10	maximum	maximum	JJ
work_55utqx7tjrft5ojtbr67ypjdye	237	11	p	p	NN
work_55utqx7tjrft5ojtbr67ypjdye	237	12	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	237	13	0.01	0.01	CD
work_55utqx7tjrft5ojtbr67ypjdye	237	14	value	value	NN
work_55utqx7tjrft5ojtbr67ypjdye	237	15	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	237	16	any	any	DT
work_55utqx7tjrft5ojtbr67ypjdye	237	17	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	237	18	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	237	19	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	237	20	compared	compare	VBN
work_55utqx7tjrft5ojtbr67ypjdye	237	21	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	237	22	itself	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	237	23	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	237	24	i.e.	i.e.	FW
work_55utqx7tjrft5ojtbr67ypjdye	238	1	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	238	2	95	95	CD
work_55utqx7tjrft5ojtbr67ypjdye	238	3	%	%	NN
work_55utqx7tjrft5ojtbr67ypjdye	238	4	confidence	confidence	NN
work_55utqx7tjrft5ojtbr67ypjdye	238	5	interval	interval	NN
work_55utqx7tjrft5ojtbr67ypjdye	238	6	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	238	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	238	8	diagonal	diagonal	JJ
work_55utqx7tjrft5ojtbr67ypjdye	238	9	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	238	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	239	1	296	296	CD
work_55utqx7tjrft5ojtbr67ypjdye	239	2	els	el	NNS
work_55utqx7tjrft5ojtbr67ypjdye	239	3	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	239	4	fewer	few	JJR
work_55utqx7tjrft5ojtbr67ypjdye	239	5	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	239	6	over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	239	7	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	239	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	240	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	240	2	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	240	3	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	240	4	im-	im-	VB
work_55utqx7tjrft5ojtbr67ypjdye	240	5	ply	ply	RB
work_55utqx7tjrft5ojtbr67ypjdye	240	6	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	240	7	five	five	CD
work_55utqx7tjrft5ojtbr67ypjdye	240	8	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	240	9	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	240	10	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	240	11	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	240	12	competitive	competitive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	240	13	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	240	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	240	15	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	240	16	rather	rather	RB
work_55utqx7tjrft5ojtbr67ypjdye	240	17	illustrates	illustrate	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	240	18	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	240	19	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	240	20	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	240	21	measure	measure	NN
work_55utqx7tjrft5ojtbr67ypjdye	240	22	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	240	23	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	240	24	mers	mer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	240	25	perform	perform	VBP
work_55utqx7tjrft5ojtbr67ypjdye	240	26	worse	bad	JJR
work_55utqx7tjrft5ojtbr67ypjdye	240	27	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	240	28	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	240	29	naive	naive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	240	30	baseline	baseline	NN
work_55utqx7tjrft5ojtbr67ypjdye	240	31	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	240	32	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	240	33	cor-	cor-	NN
work_55utqx7tjrft5ojtbr67ypjdye	240	34	pus	pus	NN
work_55utqx7tjrft5ojtbr67ypjdye	240	35	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	240	36	short	short	JJ
work_55utqx7tjrft5ojtbr67ypjdye	240	37	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	240	38	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	240	39	irregular	irregular	JJ
work_55utqx7tjrft5ojtbr67ypjdye	240	40	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	240	41	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	241	1	5.5	5.5	CD
work_55utqx7tjrft5ojtbr67ypjdye	241	2	Influential	influential	JJ
work_55utqx7tjrft5ojtbr67ypjdye	241	3	Words	Words	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	241	4	To	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	241	5	identify	identify	VB
work_55utqx7tjrft5ojtbr67ypjdye	241	6	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	241	7	types	type	NNS
work_55utqx7tjrft5ojtbr67ypjdye	241	8	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	241	9	positively	positively	RB
work_55utqx7tjrft5ojtbr67ypjdye	241	10	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	241	11	negatively	negatively	RB
work_55utqx7tjrft5ojtbr67ypjdye	241	12	affect	affect	VB
work_55utqx7tjrft5ojtbr67ypjdye	241	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	241	14	quality	quality	NN
work_55utqx7tjrft5ojtbr67ypjdye	241	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	241	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	241	17	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	241	18	after	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	241	19	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	241	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	241	21	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	241	22	use	use	VBP
work_55utqx7tjrft5ojtbr67ypjdye	241	23	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	241	24	idf	idf	NN
work_55utqx7tjrft5ojtbr67ypjdye	241	25	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	241	26	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	241	27	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	241	28	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	241	29	metrics	metric	NNS
work_55utqx7tjrft5ojtbr67ypjdye	241	30	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	241	31	each	each	DT
work_55utqx7tjrft5ojtbr67ypjdye	241	32	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	241	33	type	type	NN
work_55utqx7tjrft5ojtbr67ypjdye	241	34	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	242	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	242	2	idf	idf	NN
work_55utqx7tjrft5ojtbr67ypjdye	242	3	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	242	4	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	242	5	metric	metric	NN
work_55utqx7tjrft5ojtbr67ypjdye	242	6	strongly	strongly	RB
work_55utqx7tjrft5ojtbr67ypjdye	242	7	indi-	indi-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	242	8	cates	cat	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	242	9	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	242	10	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	242	11	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	242	12	improves	improve	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	242	13	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	242	14	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	242	15	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	242	16	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	242	17	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	242	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	242	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	242	20	improvement	improvement	NN
work_55utqx7tjrft5ojtbr67ypjdye	242	21	applied	apply	VBD
work_55utqx7tjrft5ojtbr67ypjdye	242	22	primar-	primar-	XX
work_55utqx7tjrft5ojtbr67ypjdye	242	23	ily	ily	NNP
work_55utqx7tjrft5ojtbr67ypjdye	242	24	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	242	25	conflated	conflate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	242	26	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	242	27	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	243	1	Untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	243	2	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	243	3	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	243	4	do	do	VBP
work_55utqx7tjrft5ojtbr67ypjdye	243	5	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	243	6	share	share	VB
work_55utqx7tjrft5ojtbr67ypjdye	243	7	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	243	8	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	243	9	class	class	NN
work_55utqx7tjrft5ojtbr67ypjdye	243	10	under	under	IN
work_55utqx7tjrft5ojtbr67ypjdye	243	11	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	243	12	treatment	treatment	NN
work_55utqx7tjrft5ojtbr67ypjdye	243	13	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	243	14	e.g.	e.g.	RB
work_55utqx7tjrft5ojtbr67ypjdye	244	1	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	244	2	mar-	mar-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	244	3	quess	quess	NNP
work_55utqx7tjrft5ojtbr67ypjdye	244	4	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	244	5	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	244	6	often	often	RB
work_55utqx7tjrft5ojtbr67ypjdye	244	7	become	become	VBP
work_55utqx7tjrft5ojtbr67ypjdye	244	8	less	less	RBR
work_55utqx7tjrft5ojtbr67ypjdye	244	9	probable	probable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	244	10	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	244	11	average	average	JJ
work_55utqx7tjrft5ojtbr67ypjdye	244	12	after	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	244	13	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	244	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	245	1	Their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	245	2	inferred	inferred	JJ
work_55utqx7tjrft5ojtbr67ypjdye	245	3	hyperparameters	hyperparameter	NNS
work_55utqx7tjrft5ojtbr67ypjdye	245	4	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	245	5	larger	large	JJR
work_55utqx7tjrft5ojtbr67ypjdye	245	6	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	245	7	thus	thus	RB
work_55utqx7tjrft5ojtbr67ypjdye	245	8	encourage	encourage	VB
work_55utqx7tjrft5ojtbr67ypjdye	245	9	less	less	JJR
work_55utqx7tjrft5ojtbr67ypjdye	245	10	sparsity	sparsity	NN
work_55utqx7tjrft5ojtbr67ypjdye	245	11	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	245	12	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	245	13	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	245	14	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	245	15	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	245	16	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	245	17	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	245	18	result	result	NN
work_55utqx7tjrft5ojtbr67ypjdye	245	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	245	20	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	245	21	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	245	22	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	245	23	rarer	rare	JJR
work_55utqx7tjrft5ojtbr67ypjdye	245	24	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	245	25	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	245	26	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	245	27	own	own	JJ
work_55utqx7tjrft5ojtbr67ypjdye	245	28	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	245	29	classes	class	NNS
work_55utqx7tjrft5ojtbr67ypjdye	245	30	decreases	decrease	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	245	31	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	245	32	that	that	DT
work_55utqx7tjrft5ojtbr67ypjdye	245	33	prob-	prob-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	245	34	ability	ability	NN
work_55utqx7tjrft5ojtbr67ypjdye	245	35	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	245	36	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	245	37	distributed	distribute	VBN
work_55utqx7tjrft5ojtbr67ypjdye	245	38	across	across	IN
work_55utqx7tjrft5ojtbr67ypjdye	245	39	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	245	40	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	246	1	This	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	246	2	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	246	3	increases	increase	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	246	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	246	5	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	246	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	246	7	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	246	8	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	246	9	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	246	10	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	246	11	size-	size-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	246	12	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	246	13	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	246	14	class	class	NN
work_55utqx7tjrft5ojtbr67ypjdye	246	15	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	247	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	247	2	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	247	3	confirm	confirm	VB
work_55utqx7tjrft5ojtbr67ypjdye	247	4	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	247	5	hypotheses	hypothesis	NNS
work_55utqx7tjrft5ojtbr67ypjdye	247	6	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	247	7	earlier	early	RBR
work_55utqx7tjrft5ojtbr67ypjdye	247	8	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	247	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	247	10	paper	paper	NN
work_55utqx7tjrft5ojtbr67ypjdye	247	11	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	247	12	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	247	13	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	247	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	248	1	For	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	248	2	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	248	3	dif-	dif-	NN
work_55utqx7tjrft5ojtbr67ypjdye	248	4	ferences	ference	NNS
work_55utqx7tjrft5ojtbr67ypjdye	248	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	248	6	those	those	DT
work_55utqx7tjrft5ojtbr67ypjdye	248	7	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	248	8	classes	class	NNS
work_55utqx7tjrft5ojtbr67ypjdye	248	9	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	248	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	248	11	greatest	great	JJS
work_55utqx7tjrft5ojtbr67ypjdye	248	12	weighted	weighted	JJ
work_55utqx7tjrft5ojtbr67ypjdye	248	13	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	248	14	improvement	improvement	NN
work_55utqx7tjrft5ojtbr67ypjdye	248	15	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	248	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	248	17	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	248	18	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	248	19	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	248	20	ArXiv	ArXiv	NNP
work_55utqx7tjrft5ojtbr67ypjdye	248	21	include	include	VBP
work_55utqx7tjrft5ojtbr67ypjdye	248	22	huge	huge	JJ
work_55utqx7tjrft5ojtbr67ypjdye	248	23	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	248	24	classes	class	NNS
work_55utqx7tjrft5ojtbr67ypjdye	248	25	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	248	26	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	248	27	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	248	28	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	248	29	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	248	30	prefix	prefix	NN
work_55utqx7tjrft5ojtbr67ypjdye	248	31	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	248	32	wildly	wildly	RB
work_55utqx7tjrft5ojtbr67ypjdye	248	33	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	248	34	roots	root	NNS
work_55utqx7tjrft5ojtbr67ypjdye	248	35	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	249	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	249	2	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	249	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	249	4	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	249	5	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	249	6	forced	force	VBN
work_55utqx7tjrft5ojtbr67ypjdye	249	7	sparsity	sparsity	NN
work_55utqx7tjrft5ojtbr67ypjdye	249	8	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	249	9	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	249	10	should	should	MD
work_55utqx7tjrft5ojtbr67ypjdye	249	11	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	249	12	necessarily	necessarily	RB
work_55utqx7tjrft5ojtbr67ypjdye	249	13	have	have	VB
work_55utqx7tjrft5ojtbr67ypjdye	249	14	been	be	VBN
work_55utqx7tjrft5ojtbr67ypjdye	249	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	249	16	degrading	degrade	VBG
work_55utqx7tjrft5ojtbr67ypjdye	249	17	coher-	coher-	NN
work_55utqx7tjrft5ojtbr67ypjdye	249	18	ence	ence	NN
work_55utqx7tjrft5ojtbr67ypjdye	249	19	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	250	1	As	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	250	2	exemplified	exemplify	VBN
work_55utqx7tjrft5ojtbr67ypjdye	250	3	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	250	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	250	5	50-topic	50-topic	CD
work_55utqx7tjrft5ojtbr67ypjdye	250	6	NYT	NYT	NNP
work_55utqx7tjrft5ojtbr67ypjdye	250	7	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	250	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	250	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	250	10	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	250	11	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	12	improves	improve	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	250	13	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	250	14	likelihood	likelihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	250	16	com-	com-	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	17	mon	mon	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	18	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	250	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	250	20	like	like	IN
work_55utqx7tjrft5ojtbr67ypjdye	250	21	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	250	22	street	street	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	23	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	250	24	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	250	25	TPscore	tpscore	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	26	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	250	27	5370	5370	CD
work_55utqx7tjrft5ojtbr67ypjdye	250	28	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	250	29	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	250	30	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	250	31	mr	mr	NNP
work_55utqx7tjrft5ojtbr67ypjdye	250	32	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	250	33	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	250	34	TPscore	tpscore	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	35	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	250	36	13945	13945	CD
work_55utqx7tjrft5ojtbr67ypjdye	250	37	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	250	38	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	250	39	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	250	40	outcome	outcome	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	41	aligned	align	VBN
work_55utqx7tjrft5ojtbr67ypjdye	250	42	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	250	43	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	250	44	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	45	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	250	46	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	250	47	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	48	’s	’s	POS
work_55utqx7tjrft5ojtbr67ypjdye	250	49	aim	aim	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	50	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	250	51	cope	cope	VB
work_55utqx7tjrft5ojtbr67ypjdye	250	52	well	well	RB
work_55utqx7tjrft5ojtbr67ypjdye	250	53	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	250	54	com-	com-	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	55	mon	mon	NN
work_55utqx7tjrft5ojtbr67ypjdye	250	56	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	250	57	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	251	1	But	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	251	2	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	251	3	rarer	rare	JJR
work_55utqx7tjrft5ojtbr67ypjdye	251	4	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	251	5	like	like	IN
work_55utqx7tjrft5ojtbr67ypjdye	251	6	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	251	7	purgative	purgative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	251	8	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	251	9	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	251	10	TPscore	tpscore	NN
work_55utqx7tjrft5ojtbr67ypjdye	251	11	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	251	12	−17.5	−17.5	NN
work_55utqx7tjrft5ojtbr67ypjdye	251	13	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	251	14	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	251	15	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	251	16	pranks	prank	NNS
work_55utqx7tjrft5ojtbr67ypjdye	251	17	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	251	18	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	251	19	TPscore	tpscore	NN
work_55utqx7tjrft5ojtbr67ypjdye	251	20	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	251	21	−15.4	−15.4	NNP
work_55utqx7tjrft5ojtbr67ypjdye	251	22	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	251	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	251	24	no	no	DT
work_55utqx7tjrft5ojtbr67ypjdye	251	25	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	251	26	improvement	improvement	NN
work_55utqx7tjrft5ojtbr67ypjdye	251	27	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	251	28	seen	see	VBN
work_55utqx7tjrft5ojtbr67ypjdye	251	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	252	1	These	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	252	2	com-	com-	NN
work_55utqx7tjrft5ojtbr67ypjdye	252	3	mon	mon	NN
work_55utqx7tjrft5ojtbr67ypjdye	252	4	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	252	5	do	do	VBP
work_55utqx7tjrft5ojtbr67ypjdye	252	6	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	252	7	have	have	VB
work_55utqx7tjrft5ojtbr67ypjdye	252	8	extreme	extreme	JJ
work_55utqx7tjrft5ojtbr67ypjdye	252	9	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	252	10	values	value	NNS
work_55utqx7tjrft5ojtbr67ypjdye	252	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	252	12	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	252	13	supports	support	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	252	14	our	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	252	15	hypothesis	hypothesis	NN
work_55utqx7tjrft5ojtbr67ypjdye	252	16	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	252	17	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	252	18	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	252	19	likeli-	likeli-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	252	20	hood	hood	NN
work_55utqx7tjrft5ojtbr67ypjdye	252	21	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	252	22	common	common	JJ
work_55utqx7tjrft5ojtbr67ypjdye	252	23	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	252	24	improves	improve	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	252	25	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	252	26	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	252	27	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	252	28	ming	ming	NNP
work_55utqx7tjrft5ojtbr67ypjdye	252	29	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	252	30	those	those	DT
work_55utqx7tjrft5ojtbr67ypjdye	252	31	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	252	32	were	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	252	33	already	already	RB
work_55utqx7tjrft5ojtbr67ypjdye	252	34	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	252	35	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	252	36	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	252	37	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	252	38	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	252	39	did	do	VBD
work_55utqx7tjrft5ojtbr67ypjdye	252	40	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	252	41	affect	affect	VB
work_55utqx7tjrft5ojtbr67ypjdye	252	42	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	252	43	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	252	44	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	253	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	253	2	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	253	3	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	253	4	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	253	5	use	use	VB
work_55utqx7tjrft5ojtbr67ypjdye	253	6	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	253	7	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	253	8	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	253	9	measurement	measurement	NN
work_55utqx7tjrft5ojtbr67ypjdye	253	10	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	253	11	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	253	12	context	context	NN
work_55utqx7tjrft5ojtbr67ypjdye	253	13	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	253	14	sensitive	sensitive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	253	15	lemmatizer	lemmatizer	NN
work_55utqx7tjrft5ojtbr67ypjdye	253	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	253	17	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	253	18	see	see	VBP
work_55utqx7tjrft5ojtbr67ypjdye	253	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	253	20	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	253	21	ef-	ef-	XX
work_55utqx7tjrft5ojtbr67ypjdye	253	22	fect	fect	NN
work_55utqx7tjrft5ojtbr67ypjdye	253	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	253	24	where	where	WRB
work_55utqx7tjrft5ojtbr67ypjdye	253	25	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	253	26	most	most	RBS
work_55utqx7tjrft5ojtbr67ypjdye	253	27	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	253	28	improved	improve	VBN
work_55utqx7tjrft5ojtbr67ypjdye	253	29	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	253	30	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	253	31	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	253	32	most	most	RBS
work_55utqx7tjrft5ojtbr67ypjdye	253	33	common	common	JJ
work_55utqx7tjrft5ojtbr67ypjdye	253	34	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	253	35	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	253	36	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	253	37	less	less	RBR
work_55utqx7tjrft5ojtbr67ypjdye	253	38	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	253	39	likely	likely	JJ
work_55utqx7tjrft5ojtbr67ypjdye	253	40	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	253	41	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	253	42	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	253	43	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	253	44	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	253	45	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	253	46	rare	rare	JJ
work_55utqx7tjrft5ojtbr67ypjdye	253	47	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	253	48	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	253	49	names	name	NNS
work_55utqx7tjrft5ojtbr67ypjdye	253	50	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	254	1	Interesting	interesting	JJ
work_55utqx7tjrft5ojtbr67ypjdye	254	2	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	254	3	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	254	4	arise	arise	VBP
work_55utqx7tjrft5ojtbr67ypjdye	254	5	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	254	6	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	254	7	five-	five-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	254	8	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	254	9	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	254	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	255	1	Unlike	unlike	IN
work_55utqx7tjrft5ojtbr67ypjdye	255	2	prescriptive	prescriptive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	255	3	rule	rule	NN
work_55utqx7tjrft5ojtbr67ypjdye	255	4	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	255	5	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	255	6	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	255	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	255	8	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	255	9	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	255	10	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	255	11	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	255	12	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	255	13	produce	produce	VB
work_55utqx7tjrft5ojtbr67ypjdye	255	14	more	more	JJR
work_55utqx7tjrft5ojtbr67ypjdye	255	15	errors	error	NNS
work_55utqx7tjrft5ojtbr67ypjdye	255	16	when	when	WRB
work_55utqx7tjrft5ojtbr67ypjdye	255	17	typos	typos	NNP
work_55utqx7tjrft5ojtbr67ypjdye	255	18	arise	arise	VBP
work_55utqx7tjrft5ojtbr67ypjdye	255	19	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	255	20	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	255	21	fact	fact	NN
work_55utqx7tjrft5ojtbr67ypjdye	255	22	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	255	23	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	255	24	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	255	25	ac-	ac-	RB
work_55utqx7tjrft5ojtbr67ypjdye	255	26	commodate	commodate	VB
work_55utqx7tjrft5ojtbr67ypjdye	255	27	typos	typo	NNS
work_55utqx7tjrft5ojtbr67ypjdye	255	28	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	255	29	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	255	30	ends	end	NNS
work_55utqx7tjrft5ojtbr67ypjdye	255	31	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	255	32	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	255	33	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	255	34	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	255	35	way	way	NN
work_55utqx7tjrft5ojtbr67ypjdye	255	36	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	255	37	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	255	38	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	255	39	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	255	40	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	255	41	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	256	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	256	3	once	once	RB
work_55utqx7tjrft5ojtbr67ypjdye	256	4	again	again	RB
work_55utqx7tjrft5ojtbr67ypjdye	256	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	256	6	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	256	7	observe	observe	VBP
work_55utqx7tjrft5ojtbr67ypjdye	256	8	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	256	10	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	256	11	probabilities	probability	NNS
work_55utqx7tjrft5ojtbr67ypjdye	256	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	13	truncated	truncate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	256	14	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	256	15	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	256	16	much	much	RB
work_55utqx7tjrft5ojtbr67ypjdye	256	17	improved	improve	VBN
work_55utqx7tjrft5ojtbr67ypjdye	256	18	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	19	common	common	JJ
work_55utqx7tjrft5ojtbr67ypjdye	256	20	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	256	21	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	256	22	slightly	slightly	RB
work_55utqx7tjrft5ojtbr67ypjdye	256	23	reduced	reduce	VBN
work_55utqx7tjrft5ojtbr67ypjdye	256	24	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	25	rare	rare	JJ
work_55utqx7tjrft5ojtbr67ypjdye	256	26	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	256	27	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	256	28	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	256	29	discover	discover	VBP
work_55utqx7tjrft5ojtbr67ypjdye	256	30	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	31	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	256	32	best	good	JJS
work_55utqx7tjrft5ojtbr67ypjdye	256	33	entropy	entropy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	256	34	improvements	improvement	NNS
work_55utqx7tjrft5ojtbr67ypjdye	256	35	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	36	untreated	untreated	JJ
work_55utqx7tjrft5ojtbr67ypjdye	256	37	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	38	stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	256	39	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	256	40	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	256	41	elongated	elongate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	256	42	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	256	43	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	256	44	exclama-	exclama-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	256	45	tions	tion	NNS
work_55utqx7tjrft5ojtbr67ypjdye	256	46	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	256	47	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	256	48	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	256	49	eeeee	eeeee	NNP
work_55utqx7tjrft5ojtbr67ypjdye	256	50	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	256	51	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	256	52	∆Hw(k	∆Hw(k	NNP
work_55utqx7tjrft5ojtbr67ypjdye	256	53	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	256	54	=	=	NFP
work_55utqx7tjrft5ojtbr67ypjdye	256	55	−2.56	−2.56	NNP
work_55utqx7tjrft5ojtbr67ypjdye	256	56	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	256	57	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	256	58	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	256	59	haaaa	haaaa	FW
work_55utqx7tjrft5ojtbr67ypjdye	256	60	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	256	61	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	256	62	∆Hw(k	∆hw(k	RB
work_55utqx7tjrft5ojtbr67ypjdye	256	63	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	256	64	=	=	SYM
work_55utqx7tjrft5ojtbr67ypjdye	256	65	−3.25	−3.25	NNP
work_55utqx7tjrft5ojtbr67ypjdye	256	66	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	256	67	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	257	1	At	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	257	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	257	3	opposite	opposite	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	4	score	score	NN
work_55utqx7tjrft5ojtbr67ypjdye	257	5	extreme	extreme	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	257	7	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	8	classes	class	NNS
work_55utqx7tjrft5ojtbr67ypjdye	257	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	257	10	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	257	11	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	257	12	many	many	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	13	mis-	mis-	NNS
work_55utqx7tjrft5ojtbr67ypjdye	257	14	spellings	spelling	NNS
work_55utqx7tjrft5ojtbr67ypjdye	257	15	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	257	16	increased	increase	VBN
work_55utqx7tjrft5ojtbr67ypjdye	257	17	entropy	entropy	RB
work_55utqx7tjrft5ojtbr67ypjdye	257	18	after	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	257	19	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	257	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	257	21	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	257	22	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	257	23	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	257	24	potentially	potentially	RB
work_55utqx7tjrft5ojtbr67ypjdye	257	25	misleading	misleading	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	26	;	;	:
work_55utqx7tjrft5ojtbr67ypjdye	257	27	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	257	28	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	257	29	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	257	30	very	very	RB
work_55utqx7tjrft5ojtbr67ypjdye	257	31	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	32	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	257	33	distinguishing	distinguish	VBG
work_55utqx7tjrft5ojtbr67ypjdye	257	34	dialects	dialect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	257	35	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	257	36	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	257	37	system-	system-	XX
work_55utqx7tjrft5ojtbr67ypjdye	257	38	atic	atic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	39	misspellings	misspelling	NNS
work_55utqx7tjrft5ojtbr67ypjdye	257	40	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	257	41	likely	likely	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	42	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	257	43	create	create	VB
work_55utqx7tjrft5ojtbr67ypjdye	257	44	differently-	differently-	NN
work_55utqx7tjrft5ojtbr67ypjdye	257	45	spelled	spell	VBN
work_55utqx7tjrft5ojtbr67ypjdye	257	46	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	257	47	semantically	semantically	RB
work_55utqx7tjrft5ojtbr67ypjdye	257	48	similar	similar	JJ
work_55utqx7tjrft5ojtbr67ypjdye	257	49	topics	topic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	257	50	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	257	51	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	257	52	many-	many-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	257	53	topic	topic	NNP
work_55utqx7tjrft5ojtbr67ypjdye	257	54	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	257	55	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	258	1	Over	over	IN
work_55utqx7tjrft5ojtbr67ypjdye	258	2	one	one	CD
work_55utqx7tjrft5ojtbr67ypjdye	258	3	hundred	hundred	CD
work_55utqx7tjrft5ojtbr67ypjdye	258	4	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	258	5	conflate	conflate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	258	6	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	258	7	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	258	8	defin	defin	VB
work_55utqx7tjrft5ojtbr67ypjdye	258	9	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	258	10	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	258	11	five	five	CD
work_55utqx7tjrft5ojtbr67ypjdye	258	12	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	258	13	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	258	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	258	15	including	include	VBG
work_55utqx7tjrft5ojtbr67ypjdye	258	16	upwards	upward	NNS
work_55utqx7tjrft5ojtbr67ypjdye	258	17	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	258	18	sixty	sixty	CD
work_55utqx7tjrft5ojtbr67ypjdye	258	19	misspellings	misspelling	NNS
work_55utqx7tjrft5ojtbr67ypjdye	258	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	258	21	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	258	22	definitely	definitely	RB
work_55utqx7tjrft5ojtbr67ypjdye	258	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	258	24	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	258	25	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	258	26	removes	remove	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	258	27	distinction	distinction	NN
work_55utqx7tjrft5ojtbr67ypjdye	258	28	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	258	29	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	258	30	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	258	31	bad	bad	JJ
work_55utqx7tjrft5ojtbr67ypjdye	258	32	spellers	speller	NNS
work_55utqx7tjrft5ojtbr67ypjdye	258	33	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	258	34	might	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	258	35	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	258	36	correlated	correlate	VBN
work_55utqx7tjrft5ojtbr67ypjdye	258	37	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	258	38	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	258	39	features	feature	NNS
work_55utqx7tjrft5ojtbr67ypjdye	258	40	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	259	1	6	6	CD
work_55utqx7tjrft5ojtbr67ypjdye	259	2	Related	related	JJ
work_55utqx7tjrft5ojtbr67ypjdye	259	3	Work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	259	4	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	259	5	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	259	6	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	259	7	aware	aware	JJ
work_55utqx7tjrft5ojtbr67ypjdye	259	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	259	9	other	other	JJ
work_55utqx7tjrft5ojtbr67ypjdye	259	10	work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	259	11	evaluating	evaluate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	259	12	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	259	13	vari-	vari-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	259	14	ety	ety	NN
work_55utqx7tjrft5ojtbr67ypjdye	259	15	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	259	16	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	259	17	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	259	18	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	259	19	techniques	technique	NNS
work_55utqx7tjrft5ojtbr67ypjdye	259	20	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	259	21	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	259	22	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	259	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	260	1	Some	some	DT
work_55utqx7tjrft5ojtbr67ypjdye	260	2	prior	prior	JJ
work_55utqx7tjrft5ojtbr67ypjdye	260	3	work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	260	4	exists	exist	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	260	5	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	260	6	evaluates	evaluate	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	260	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	260	8	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	260	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	260	10	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	260	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	261	1	Several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	261	2	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	261	3	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	261	4	were	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	261	5	tested	test	VBN
work_55utqx7tjrft5ojtbr67ypjdye	261	6	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	261	7	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	261	8	variety	variety	NN
work_55utqx7tjrft5ojtbr67ypjdye	261	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	261	10	document	document	NN
work_55utqx7tjrft5ojtbr67ypjdye	261	11	clustering	cluster	VBG
work_55utqx7tjrft5ojtbr67ypjdye	261	12	algo-	algo-	XX
work_55utqx7tjrft5ojtbr67ypjdye	261	13	rithms	rithms	NN
work_55utqx7tjrft5ojtbr67ypjdye	261	14	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	261	15	Han	Han	NNP
work_55utqx7tjrft5ojtbr67ypjdye	261	16	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	261	17	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	261	18	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	262	1	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	262	2	2012	2012	CD
work_55utqx7tjrft5ojtbr67ypjdye	262	3	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	262	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	262	5	finding	find	VBG
work_55utqx7tjrft5ojtbr67ypjdye	262	6	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	262	7	they	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	262	8	could	could	MD
work_55utqx7tjrft5ojtbr67ypjdye	262	9	reduce	reduce	VB
work_55utqx7tjrft5ojtbr67ypjdye	262	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	262	11	number	number	NN
work_55utqx7tjrft5ojtbr67ypjdye	262	12	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	262	13	features	feature	NNS
work_55utqx7tjrft5ojtbr67ypjdye	262	14	effectively	effectively	RB
work_55utqx7tjrft5ojtbr67ypjdye	262	15	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	262	16	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	262	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	262	18	correct	correct	JJ
work_55utqx7tjrft5ojtbr67ypjdye	262	19	choice	choice	NN
work_55utqx7tjrft5ojtbr67ypjdye	262	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	262	21	stemmer	stemmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	262	22	varied	varied	NNP
work_55utqx7tjrft5ojtbr67ypjdye	262	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	263	1	More	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	263	2	recently	recently	RB
work_55utqx7tjrft5ojtbr67ypjdye	263	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	263	4	Stankov	Stankov	NNP
work_55utqx7tjrft5ojtbr67ypjdye	263	5	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	263	6	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	263	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	264	1	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	264	2	2013	2013	CD
work_55utqx7tjrft5ojtbr67ypjdye	264	3	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	264	4	developed	develop	VBD
work_55utqx7tjrft5ojtbr67ypjdye	264	5	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	264	6	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	264	7	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	264	8	im-	im-	UH
work_55utqx7tjrft5ojtbr67ypjdye	264	9	prove	prove	VB
work_55utqx7tjrft5ojtbr67ypjdye	264	10	document	document	NN
work_55utqx7tjrft5ojtbr67ypjdye	264	11	clustering	cluster	VBG
work_55utqx7tjrft5ojtbr67ypjdye	264	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	265	1	Both	both	DT
work_55utqx7tjrft5ojtbr67ypjdye	265	2	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	265	3	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	265	4	techniques	technique	NNS
work_55utqx7tjrft5ojtbr67ypjdye	265	5	demonstrate	demonstrate	VBP
work_55utqx7tjrft5ojtbr67ypjdye	265	6	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	265	7	improvement	improvement	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	8	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	265	9	clustering	cluster	VBG
work_55utqx7tjrft5ojtbr67ypjdye	265	10	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	265	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	265	12	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	265	13	reduction	reduction	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	14	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	265	15	features	feature	NNS
work_55utqx7tjrft5ojtbr67ypjdye	265	16	required	require	VBN
work_55utqx7tjrft5ojtbr67ypjdye	265	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	265	18	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	265	19	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	265	20	for-	for-	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	21	mer	mer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	265	22	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	265	23	introducing	introduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	265	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	265	25	notion	notion	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	26	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	265	27	tradeoff	tradeoff	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	28	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	265	29	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	265	30	precision	precision	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	31	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	265	32	lemmatization	lemmatization	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	33	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	265	34	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	265	35	efficiency	efficiency	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	36	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	265	37	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	265	38	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	265	39	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	265	40	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	266	1	Additionally	additionally	RB
work_55utqx7tjrft5ojtbr67ypjdye	266	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	266	3	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	266	4	variety	variety	NN
work_55utqx7tjrft5ojtbr67ypjdye	266	5	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	266	6	work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	266	7	exists	exist	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	266	8	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	266	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	266	10	gen-	gen-	NN
work_55utqx7tjrft5ojtbr67ypjdye	266	11	eral	eral	JJ
work_55utqx7tjrft5ojtbr67ypjdye	266	12	field	field	NN
work_55utqx7tjrft5ojtbr67ypjdye	266	13	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	266	14	stemmer	stemmer	JJ
work_55utqx7tjrft5ojtbr67ypjdye	266	15	evaluation	evaluation	NN
work_55utqx7tjrft5ojtbr67ypjdye	266	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	266	17	though	though	IN
work_55utqx7tjrft5ojtbr67ypjdye	266	18	much	much	JJ
work_55utqx7tjrft5ojtbr67ypjdye	266	19	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	266	20	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	266	21	centers	center	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	266	22	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	266	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	266	24	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	266	25	retrieval	retrieval	NNP
work_55utqx7tjrft5ojtbr67ypjdye	266	26	community	community	NN
work_55utqx7tjrft5ojtbr67ypjdye	266	27	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	267	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	267	2	particular	particular	JJ
work_55utqx7tjrft5ojtbr67ypjdye	267	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	267	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	267	5	work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	267	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	267	7	Harman	Harman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	267	8	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	267	9	1991	1991	CD
work_55utqx7tjrft5ojtbr67ypjdye	267	10	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	267	11	highlights	highlight	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	267	12	some	some	DT
work_55utqx7tjrft5ojtbr67ypjdye	267	13	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	267	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	267	15	fundamental	fundamental	JJ
work_55utqx7tjrft5ojtbr67ypjdye	267	16	issues	issue	NNS
work_55utqx7tjrft5ojtbr67ypjdye	267	17	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	267	18	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	267	19	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	267	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	267	21	297	297	CD
work_55utqx7tjrft5ojtbr67ypjdye	267	22	including	include	VBG
work_55utqx7tjrft5ojtbr67ypjdye	267	23	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	267	24	potential	potential	JJ
work_55utqx7tjrft5ojtbr67ypjdye	267	25	positive	positive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	267	26	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	267	27	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	267	28	light	light	JJ
work_55utqx7tjrft5ojtbr67ypjdye	267	29	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	267	30	mers	mer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	267	31	like	like	IN
work_55utqx7tjrft5ojtbr67ypjdye	267	32	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	267	33	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	267	34	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	267	35	removal	removal	NN
work_55utqx7tjrft5ojtbr67ypjdye	267	36	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	267	37	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	268	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	268	2	notion	notion	NN
work_55utqx7tjrft5ojtbr67ypjdye	268	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	268	4	stemmer	stemmer	JJ
work_55utqx7tjrft5ojtbr67ypjdye	268	5	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	268	6	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	268	7	detailed	detail	VBN
work_55utqx7tjrft5ojtbr67ypjdye	268	8	further	further	RB
work_55utqx7tjrft5ojtbr67ypjdye	268	9	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	268	10	Frakes	Frakes	NNP
work_55utqx7tjrft5ojtbr67ypjdye	268	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	268	12	Fox	Fox	NNP
work_55utqx7tjrft5ojtbr67ypjdye	268	13	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	268	14	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	268	15	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	268	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	268	17	as	as	RB
work_55utqx7tjrft5ojtbr67ypjdye	268	18	well	well	RB
work_55utqx7tjrft5ojtbr67ypjdye	268	19	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	268	20	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	268	21	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	268	22	precise	precise	JJ
work_55utqx7tjrft5ojtbr67ypjdye	268	23	met-	met-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	268	24	rics	ric	NNS
work_55utqx7tjrft5ojtbr67ypjdye	268	25	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	268	26	evaluation	evaluation	NN
work_55utqx7tjrft5ojtbr67ypjdye	268	27	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	268	28	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	268	29	strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	268	30	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	269	1	Survey	survey	VB
work_55utqx7tjrft5ojtbr67ypjdye	269	2	pa-	pa-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	269	3	pers	per	NNS
work_55utqx7tjrft5ojtbr67ypjdye	269	4	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	269	5	Jivani	Jivani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	269	6	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	269	7	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	269	8	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	269	9	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	269	10	Rani	Rani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	269	11	et	et	NNP
work_55utqx7tjrft5ojtbr67ypjdye	269	12	al	al	NNP
work_55utqx7tjrft5ojtbr67ypjdye	269	13	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	270	1	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	270	2	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	270	3	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	270	4	detail	detail	VBP
work_55utqx7tjrft5ojtbr67ypjdye	270	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	270	6	different	different	JJ
work_55utqx7tjrft5ojtbr67ypjdye	270	7	existing	exist	VBG
work_55utqx7tjrft5ojtbr67ypjdye	270	8	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	270	9	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	270	10	conflation	conflation	NN
work_55utqx7tjrft5ojtbr67ypjdye	270	11	tech-	tech-	NN
work_55utqx7tjrft5ojtbr67ypjdye	270	12	niques	nique	NNS
work_55utqx7tjrft5ojtbr67ypjdye	270	13	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	270	14	machine	machine	NN
work_55utqx7tjrft5ojtbr67ypjdye	270	15	learning	learning	NN
work_55utqx7tjrft5ojtbr67ypjdye	270	16	applications	application	NNS
work_55utqx7tjrft5ojtbr67ypjdye	270	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	270	18	including	include	VBG
work_55utqx7tjrft5ojtbr67ypjdye	270	19	several	several	JJ
work_55utqx7tjrft5ojtbr67ypjdye	270	20	statistical	statistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	270	21	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	270	22	algorithms	algorithm	NNS
work_55utqx7tjrft5ojtbr67ypjdye	270	23	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	270	24	do	do	VBP
work_55utqx7tjrft5ojtbr67ypjdye	270	25	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	270	26	rely	rely	VB
work_55utqx7tjrft5ojtbr67ypjdye	270	27	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	270	28	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	270	29	fixed	fix	VBN
work_55utqx7tjrft5ojtbr67ypjdye	270	30	set	set	NN
work_55utqx7tjrft5ojtbr67ypjdye	270	31	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	270	32	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	270	33	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	271	1	Findings	finding	NNS
work_55utqx7tjrft5ojtbr67ypjdye	271	2	suggest	suggest	VBP
work_55utqx7tjrft5ojtbr67ypjdye	271	3	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	271	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	271	5	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	271	6	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	271	7	statistical	statistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	271	8	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	271	9	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	271	10	potential	potential	NN
work_55utqx7tjrft5ojtbr67ypjdye	271	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	271	12	many	many	JJ
work_55utqx7tjrft5ojtbr67ypjdye	271	13	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	271	14	inefficient	inefficient	JJ
work_55utqx7tjrft5ojtbr67ypjdye	271	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	271	16	complex	complex	JJ
work_55utqx7tjrft5ojtbr67ypjdye	271	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	271	18	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	271	19	difficult	difficult	JJ
work_55utqx7tjrft5ojtbr67ypjdye	271	20	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	271	21	calibrate	calibrate	VB
work_55utqx7tjrft5ojtbr67ypjdye	271	22	well	well	RB
work_55utqx7tjrft5ojtbr67ypjdye	271	23	enough	enough	JJ
work_55utqx7tjrft5ojtbr67ypjdye	271	24	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	271	25	produce	produce	VB
work_55utqx7tjrft5ojtbr67ypjdye	271	26	good	good	JJ
work_55utqx7tjrft5ojtbr67ypjdye	271	27	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	271	28	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	272	1	Though	though	IN
work_55utqx7tjrft5ojtbr67ypjdye	272	2	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	272	3	look	look	VBP
work_55utqx7tjrft5ojtbr67ypjdye	272	4	forward	forward	RB
work_55utqx7tjrft5ojtbr67ypjdye	272	5	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	272	6	seeing	see	VBG
work_55utqx7tjrft5ojtbr67ypjdye	272	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	272	8	future	future	JJ
work_55utqx7tjrft5ojtbr67ypjdye	272	9	development	development	NN
work_55utqx7tjrft5ojtbr67ypjdye	272	10	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	272	11	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	272	12	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	272	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	272	14	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	272	15	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	272	16	work	work	NN
work_55utqx7tjrft5ojtbr67ypjdye	272	17	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	272	18	chose	choose	VBD
work_55utqx7tjrft5ojtbr67ypjdye	272	19	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	272	20	focus	focus	VB
work_55utqx7tjrft5ojtbr67ypjdye	272	21	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	272	22	simpler	simple	JJR
work_55utqx7tjrft5ojtbr67ypjdye	272	23	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	272	24	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	272	25	widely	widely	RB
work_55utqx7tjrft5ojtbr67ypjdye	272	26	used	use	VBN
work_55utqx7tjrft5ojtbr67ypjdye	272	27	methods	method	NNS
work_55utqx7tjrft5ojtbr67ypjdye	272	28	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	273	1	7	7	CD
work_55utqx7tjrft5ojtbr67ypjdye	273	2	Conclusion	Conclusion	NNP
work_55utqx7tjrft5ojtbr67ypjdye	273	3	Despite	despite	IN
work_55utqx7tjrft5ojtbr67ypjdye	273	4	its	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	273	5	abiding	abide	VBG
work_55utqx7tjrft5ojtbr67ypjdye	273	6	popularity	popularity	NN
work_55utqx7tjrft5ojtbr67ypjdye	273	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	273	8	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	273	9	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	273	10	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	273	11	improve	improve	VB
work_55utqx7tjrft5ojtbr67ypjdye	273	12	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	273	13	after	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	273	14	controlling	control	VBG
work_55utqx7tjrft5ojtbr67ypjdye	273	15	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	273	16	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	273	17	size	size	NN
work_55utqx7tjrft5ojtbr67ypjdye	273	18	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	273	19	vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	273	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	273	21	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	273	22	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	273	23	actually	actually	RB
work_55utqx7tjrft5ojtbr67ypjdye	273	24	reduce	reduce	VB
work_55utqx7tjrft5ojtbr67ypjdye	273	25	predictive	predictive	JJ
work_55utqx7tjrft5ojtbr67ypjdye	273	26	like-	like-	NN
work_55utqx7tjrft5ojtbr67ypjdye	273	27	lihood	lihood	NN
work_55utqx7tjrft5ojtbr67ypjdye	273	28	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	273	29	increase	increase	VB
work_55utqx7tjrft5ojtbr67ypjdye	273	30	sensitivity	sensitivity	NN
work_55utqx7tjrft5ojtbr67ypjdye	273	31	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	273	32	random	random	JJ
work_55utqx7tjrft5ojtbr67ypjdye	273	33	initializa-	initializa-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	273	34	tions	tion	NNS
work_55utqx7tjrft5ojtbr67ypjdye	273	35	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	274	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	274	2	most	most	JJS
work_55utqx7tjrft5ojtbr67ypjdye	274	3	cases	case	NNS
work_55utqx7tjrft5ojtbr67ypjdye	274	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	274	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	274	6	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	274	7	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	274	8	was	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	274	9	already	already	RB
work_55utqx7tjrft5ojtbr67ypjdye	274	10	grouping	group	VBG
work_55utqx7tjrft5ojtbr67ypjdye	274	11	together	together	RB
work_55utqx7tjrft5ojtbr67ypjdye	274	12	common	common	JJ
work_55utqx7tjrft5ojtbr67ypjdye	274	13	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	274	14	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	274	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	274	16	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	274	17	root	root	NN
work_55utqx7tjrft5ojtbr67ypjdye	274	18	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	274	19	its	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	274	20	own	own	JJ
work_55utqx7tjrft5ojtbr67ypjdye	274	21	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	274	22	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	274	23	gained	gain	VBN
work_55utqx7tjrft5ojtbr67ypjdye	274	24	little	little	JJ
work_55utqx7tjrft5ojtbr67ypjdye	274	25	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	274	26	better	well	JJR
work_55utqx7tjrft5ojtbr67ypjdye	274	27	model-	model-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	274	28	ing	e	VBG
work_55utqx7tjrft5ojtbr67ypjdye	274	29	rare	rare	JJ
work_55utqx7tjrft5ojtbr67ypjdye	274	30	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	274	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	275	1	Light	light	JJ
work_55utqx7tjrft5ojtbr67ypjdye	275	2	treatments	treatment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	275	3	seem	seem	VBP
work_55utqx7tjrft5ojtbr67ypjdye	275	4	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	275	5	fare	fare	VB
work_55utqx7tjrft5ojtbr67ypjdye	275	6	better	well	RBR
work_55utqx7tjrft5ojtbr67ypjdye	275	7	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	275	8	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	275	9	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	275	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	275	11	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	275	12	Krovetz	Krovetz	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	275	13	doing	do	VBG
work_55utqx7tjrft5ojtbr67ypjdye	275	14	particu-	particu-	XX
work_55utqx7tjrft5ojtbr67ypjdye	275	15	larly	larly	RB
work_55utqx7tjrft5ojtbr67ypjdye	275	16	well	well	RB
work_55utqx7tjrft5ojtbr67ypjdye	275	17	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	275	18	well	well	RB
work_55utqx7tjrft5ojtbr67ypjdye	275	19	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	275	20	proofread	proofread	NN
work_55utqx7tjrft5ojtbr67ypjdye	275	21	corpora	corpora	NN
work_55utqx7tjrft5ojtbr67ypjdye	275	22	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	275	23	but	but	CC
work_55utqx7tjrft5ojtbr67ypjdye	275	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	275	25	small	small	JJ
work_55utqx7tjrft5ojtbr67ypjdye	275	26	differences	difference	NNS
work_55utqx7tjrft5ojtbr67ypjdye	275	27	between	between	IN
work_55utqx7tjrft5ojtbr67ypjdye	275	28	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	275	29	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	275	30	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	275	31	target	target	VBP
work_55utqx7tjrft5ojtbr67ypjdye	275	32	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	275	33	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	275	34	pluralization	pluralization	NN
work_55utqx7tjrft5ojtbr67ypjdye	275	35	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	275	36	verb	verb	JJ
work_55utqx7tjrft5ojtbr67ypjdye	275	37	conjugation	conjugation	NN
work_55utqx7tjrft5ojtbr67ypjdye	275	38	are	be	VBP
work_55utqx7tjrft5ojtbr67ypjdye	275	39	often	often	RB
work_55utqx7tjrft5ojtbr67ypjdye	275	40	already	already	RB
work_55utqx7tjrft5ojtbr67ypjdye	275	41	captured	capture	VBN
work_55utqx7tjrft5ojtbr67ypjdye	275	42	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	275	43	semantic	semantic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	275	44	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	275	45	like	like	IN
work_55utqx7tjrft5ojtbr67ypjdye	275	46	LDA	LDA	NNP
work_55utqx7tjrft5ojtbr67ypjdye	275	47	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	276	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	276	2	certain	certain	JJ
work_55utqx7tjrft5ojtbr67ypjdye	276	3	cases	case	NNS
work_55utqx7tjrft5ojtbr67ypjdye	276	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	276	5	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	276	6	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	276	7	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	276	8	encode	encode	VB
work_55utqx7tjrft5ojtbr67ypjdye	276	9	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	276	10	as-	as-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	276	11	sumption	sumption	NN
work_55utqx7tjrft5ojtbr67ypjdye	276	12	that	that	WDT
work_55utqx7tjrft5ojtbr67ypjdye	276	13	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	276	14	useful	useful	JJ
work_55utqx7tjrft5ojtbr67ypjdye	276	15	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	276	16	coping	cope	VBG
work_55utqx7tjrft5ojtbr67ypjdye	276	17	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	276	18	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	276	19	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	276	20	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	276	21	heavy	heavy	JJ
work_55utqx7tjrft5ojtbr67ypjdye	276	22	variation	variation	NN
work_55utqx7tjrft5ojtbr67ypjdye	276	23	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	276	24	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	276	25	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	276	26	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	276	27	5-truncation	5-truncation	CD
work_55utqx7tjrft5ojtbr67ypjdye	276	28	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	276	29	helping	helping	NN
work_55utqx7tjrft5ojtbr67ypjdye	276	30	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	276	31	correct	correct	VB
work_55utqx7tjrft5ojtbr67ypjdye	276	32	misspellings	misspelling	NNS
work_55utqx7tjrft5ojtbr67ypjdye	276	33	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	276	34	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	276	35	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	277	1	While	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	277	2	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	277	3	does	do	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	277	4	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	277	5	improve	improve	VB
work_55utqx7tjrft5ojtbr67ypjdye	277	6	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	277	7	quality	quality	NN
work_55utqx7tjrft5ojtbr67ypjdye	277	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	277	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	277	10	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	277	11	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	277	12	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	277	13	most	most	JJS
work_55utqx7tjrft5ojtbr67ypjdye	277	14	measures	measure	NNS
work_55utqx7tjrft5ojtbr67ypjdye	277	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	277	16	it	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	277	17	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	277	18	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	277	19	suited	suit	VBN
work_55utqx7tjrft5ojtbr67ypjdye	277	20	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	277	21	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	277	22	particular	particular	JJ
work_55utqx7tjrft5ojtbr67ypjdye	277	23	task	task	NN
work_55utqx7tjrft5ojtbr67ypjdye	277	24	involving	involve	VBG
work_55utqx7tjrft5ojtbr67ypjdye	277	25	abnormally	abnormally	RB
work_55utqx7tjrft5ojtbr67ypjdye	277	26	varied	varied	JJ
work_55utqx7tjrft5ojtbr67ypjdye	277	27	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	277	28	forms	form	NNS
work_55utqx7tjrft5ojtbr67ypjdye	277	29	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	277	30	which	which	WDT
work_55utqx7tjrft5ojtbr67ypjdye	277	31	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	277	32	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	277	33	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	277	34	applied	apply	VBN
work_55utqx7tjrft5ojtbr67ypjdye	277	35	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	278	1	However	however	RB
work_55utqx7tjrft5ojtbr67ypjdye	278	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	278	3	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	278	4	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	278	5	encod-	encod-	IN
work_55utqx7tjrft5ojtbr67ypjdye	278	6	ing	ing	JJ
work_55utqx7tjrft5ojtbr67ypjdye	278	7	standard	standard	JJ
work_55utqx7tjrft5ojtbr67ypjdye	278	8	rules	rule	NNS
work_55utqx7tjrft5ojtbr67ypjdye	278	9	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	278	10	spelling	spelling	NN
work_55utqx7tjrft5ojtbr67ypjdye	278	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	278	12	grammar	grammar	NN
work_55utqx7tjrft5ojtbr67ypjdye	278	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	278	14	such	such	PDT
work_55utqx7tjrft5ojtbr67ypjdye	278	15	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	278	16	benefit	benefit	NN
work_55utqx7tjrft5ojtbr67ypjdye	278	17	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	278	18	unlikely	unlikely	JJ
work_55utqx7tjrft5ojtbr67ypjdye	278	19	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	279	1	Given	give	VBN
work_55utqx7tjrft5ojtbr67ypjdye	279	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	279	3	overly	overly	RB
work_55utqx7tjrft5ojtbr67ypjdye	279	4	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	279	5	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	279	6	effects	effect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	279	7	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	279	8	truncation	truncation	NN
work_55utqx7tjrft5ojtbr67ypjdye	279	9	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	279	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	279	11	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	279	12	suggest	suggest	VBP
work_55utqx7tjrft5ojtbr67ypjdye	279	13	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	279	14	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	279	15	stem-	stem-	NN
work_55utqx7tjrft5ojtbr67ypjdye	279	16	mer	mer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	279	17	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	279	18	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	279	19	method	method	NN
work_55utqx7tjrft5ojtbr67ypjdye	279	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	279	21	discovering	discover	VBG
work_55utqx7tjrft5ojtbr67ypjdye	279	22	misspellings	misspelling	NNS
work_55utqx7tjrft5ojtbr67ypjdye	279	23	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	279	24	fix	fix	VB
work_55utqx7tjrft5ojtbr67ypjdye	279	25	instead	instead	RB
work_55utqx7tjrft5ojtbr67ypjdye	279	26	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	279	27	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	279	28	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	279	29	way	way	NN
work_55utqx7tjrft5ojtbr67ypjdye	279	30	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	279	31	repairing	repair	VBG
work_55utqx7tjrft5ojtbr67ypjdye	279	32	them	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	279	33	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	280	1	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	280	2	common	common	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	3	motivation	motivation	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	4	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	5	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	280	6	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	280	7	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	280	8	display	display	VB
work_55utqx7tjrft5ojtbr67ypjdye	280	9	more	more	JJR
work_55utqx7tjrft5ojtbr67ypjdye	280	10	succinct	succinct	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	11	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	280	12	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	13	not	not	RB
work_55utqx7tjrft5ojtbr67ypjdye	280	14	repeating	repeat	VBG
work_55utqx7tjrft5ojtbr67ypjdye	280	15	minor	minor	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	16	mor-	mor-	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	17	phological	phological	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	18	variations	variation	NNS
work_55utqx7tjrft5ojtbr67ypjdye	280	19	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	280	20	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	21	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	22	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	280	23	place	place	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	24	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	280	25	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	280	26	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	280	27	places	place	NNS
work_55utqx7tjrft5ojtbr67ypjdye	280	28	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	280	29	Unstemmed	unstemmed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	30	room	room	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	31	hotel	hotel	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	32	stay	stay	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	33	rooms	room	NNS
work_55utqx7tjrft5ojtbr67ypjdye	280	34	pool	pool	VBP
work_55utqx7tjrft5ojtbr67ypjdye	280	35	nice	nice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	36	stayed	stay	VBD
work_55utqx7tjrft5ojtbr67ypjdye	280	37	strip	strip	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	38	night	night	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	39	bed	bed	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	40	check	check	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	41	clean	clean	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	42	bathroom	bathroom	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	43	desk	desk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	44	casino	casino	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	45	vegas	vegas	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	46	free	free	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	47	front	front	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	48	resort	resort	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	49	shower	shower	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	50	Stemmed	stem	VBN
work_55utqx7tjrft5ojtbr67ypjdye	280	51	after	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	52	training	training	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	53	room	room	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	54	hotel	hotel	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	55	stai	stai	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	56	pool	pool	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	57	nice	nice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	58	strip	strip	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	59	night	night	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	60	bed	bed	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	61	check	check	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	62	clean	clean	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	63	bathroom	bathroom	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	64	desk	desk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	65	casino	casino	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	66	vega	vega	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	67	free	free	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	68	front	front	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	69	resort	resort	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	70	shower	shower	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	71	Stemmed	Stemmed	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	72	before	before	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	73	training	training	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	74	room	room	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	75	hotel	hotel	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	76	stai	stai	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	77	pool	pool	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	78	nice	nice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	79	bed	bed	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	80	check	check	VBP
work_55utqx7tjrft5ojtbr67ypjdye	280	81	strip	strip	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	82	night	night	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	83	vega	vega	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	84	suit	suit	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	85	casino	casino	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	86	clean	clean	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	87	bath-	bath-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	88	room	room	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	89	view	view	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	90	desk	desk	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	91	resort	resort	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	92	dai	dai	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	93	walk	walk	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	94	area	area	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	95	Table	table	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	96	4	4	CD
work_55utqx7tjrft5ojtbr67ypjdye	280	97	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	280	98	An	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	280	99	example	example	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	100	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	101	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	102	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	280	103	unstemmed	unstemmed	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	104	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	280	105	50-topic	50-topic	CD
work_55utqx7tjrft5ojtbr67ypjdye	280	106	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	107	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	108	redundant	redundant	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	109	keywords	keyword	NNS
work_55utqx7tjrft5ojtbr67ypjdye	280	110	demonstrates	demonstrate	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	280	111	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	112	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	280	113	after	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	114	modeling	modeling	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	115	produces	produce	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	280	116	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	280	117	same	same	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	118	appar-	appar-	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	119	ent	ent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	120	high	high	JJ
work_55utqx7tjrft5ojtbr67ypjdye	280	121	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	280	122	probability	probability	NN
work_55utqx7tjrft5ojtbr67ypjdye	280	123	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	280	124	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	280	125	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	280	126	before	before	RB
work_55utqx7tjrft5ojtbr67ypjdye	280	127	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	281	1	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	281	2	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	281	3	case	case	NN
work_55utqx7tjrft5ojtbr67ypjdye	281	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	281	5	Yelp	Yelp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	281	6	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	281	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	282	1	As	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	282	2	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	282	3	alternative	alternative	NN
work_55utqx7tjrft5ojtbr67ypjdye	282	4	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	282	5	suggest	suggest	VBP
work_55utqx7tjrft5ojtbr67ypjdye	282	6	post	post	NN
work_55utqx7tjrft5ojtbr67ypjdye	282	7	-	-	:
work_55utqx7tjrft5ojtbr67ypjdye	282	8	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	282	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	282	10	list	list	NN
work_55utqx7tjrft5ojtbr67ypjdye	282	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	282	12	keywords	keyword	NNS
work_55utqx7tjrft5ojtbr67ypjdye	282	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	282	14	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	282	15	shown	show	VBN
work_55utqx7tjrft5ojtbr67ypjdye	282	16	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	282	17	Ta-	Ta-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	282	18	ble	ble	NNP
work_55utqx7tjrft5ojtbr67ypjdye	282	19	4	4	CD
work_55utqx7tjrft5ojtbr67ypjdye	282	20	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	283	1	Stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	283	2	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	283	3	list	list	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	283	5	top	top	JJ
work_55utqx7tjrft5ojtbr67ypjdye	283	6	words	word	NNS
work_55utqx7tjrft5ojtbr67ypjdye	283	7	after	after	IN
work_55utqx7tjrft5ojtbr67ypjdye	283	8	modeling	modeling	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	9	allows	allow	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	283	10	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	11	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	283	12	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	283	13	exploit	exploit	VB
work_55utqx7tjrft5ojtbr67ypjdye	283	14	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	283	15	nuances	nuance	NNS
work_55utqx7tjrft5ojtbr67ypjdye	283	16	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	283	17	mor-	mor-	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	18	phologies	phologie	NNS
work_55utqx7tjrft5ojtbr67ypjdye	283	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	283	20	such	such	JJ
work_55utqx7tjrft5ojtbr67ypjdye	283	21	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	283	22	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	283	23	apple	apple	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	24	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	283	25	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	283	26	“	"	``
work_55utqx7tjrft5ojtbr67ypjdye	283	27	apples	apple	NNS
work_55utqx7tjrft5ojtbr67ypjdye	283	28	”	"	''
work_55utqx7tjrft5ojtbr67ypjdye	283	29	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	283	30	respect	respect	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	31	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	283	32	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	283	33	company	company	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	34	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	283	35	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	283	36	fruit	fruit	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	37	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	283	38	while	while	IN
work_55utqx7tjrft5ojtbr67ypjdye	283	39	still	still	RB
work_55utqx7tjrft5ojtbr67ypjdye	283	40	allowing	allow	VBG
work_55utqx7tjrft5ojtbr67ypjdye	283	41	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	283	42	eventual	eventual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	283	43	viewer	viewer	NN
work_55utqx7tjrft5ojtbr67ypjdye	283	44	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	283	45	browse	browse	VB
work_55utqx7tjrft5ojtbr67ypjdye	283	46	through	through	IN
work_55utqx7tjrft5ojtbr67ypjdye	283	47	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	283	48	resulting	result	VBG
work_55utqx7tjrft5ojtbr67ypjdye	283	49	concepts	concept	NNS
work_55utqx7tjrft5ojtbr67ypjdye	283	50	quickly	quickly	RB
work_55utqx7tjrft5ojtbr67ypjdye	283	51	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	284	1	Post	post	NN
work_55utqx7tjrft5ojtbr67ypjdye	284	2	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	284	3	stemming	stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	284	4	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	284	5	computationally	computationally	RB
work_55utqx7tjrft5ojtbr67ypjdye	284	6	much	much	RB
work_55utqx7tjrft5ojtbr67ypjdye	284	7	cheaper	cheap	JJR
work_55utqx7tjrft5ojtbr67ypjdye	284	8	than	than	IN
work_55utqx7tjrft5ojtbr67ypjdye	284	9	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	284	10	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	284	11	full	full	JJ
work_55utqx7tjrft5ojtbr67ypjdye	284	12	corpus	corpus	NN
work_55utqx7tjrft5ojtbr67ypjdye	284	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	284	14	requir-	requir-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	284	15	ing	ing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	284	16	only	only	RB
work_55utqx7tjrft5ojtbr67ypjdye	284	17	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	284	18	slightly	slightly	RB
work_55utqx7tjrft5ojtbr67ypjdye	284	19	longer	long	JJR
work_55utqx7tjrft5ojtbr67ypjdye	284	20	input	input	NN
work_55utqx7tjrft5ojtbr67ypjdye	284	21	list	list	NN
work_55utqx7tjrft5ojtbr67ypjdye	284	22	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	284	23	most	most	JJS
work_55utqx7tjrft5ojtbr67ypjdye	284	24	probable	probable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	284	25	terms	term	NNS
work_55utqx7tjrft5ojtbr67ypjdye	284	26	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	285	1	Because	because	IN
work_55utqx7tjrft5ojtbr67ypjdye	285	2	context	context	NN
work_55utqx7tjrft5ojtbr67ypjdye	285	3	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	285	4	unavailable	unavailable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	285	5	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	285	6	keywords	keyword	NNS
work_55utqx7tjrft5ojtbr67ypjdye	285	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	285	8	strong	strong	JJ
work_55utqx7tjrft5ojtbr67ypjdye	285	9	stemmers	stemmer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	285	10	reduce	reduce	VBP
work_55utqx7tjrft5ojtbr67ypjdye	285	11	readability	readability	NN
work_55utqx7tjrft5ojtbr67ypjdye	285	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	285	13	we	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	285	14	would	would	MD
work_55utqx7tjrft5ojtbr67ypjdye	285	15	suggest	suggest	VB
work_55utqx7tjrft5ojtbr67ypjdye	285	16	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	285	17	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	285	18	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	285	19	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	285	20	or	or	CC
work_55utqx7tjrft5ojtbr67ypjdye	285	21	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	285	22	modification	modification	NN
work_55utqx7tjrft5ojtbr67ypjdye	285	23	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	285	24	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	285	25	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	285	26	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	285	27	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	285	28	return	return	VB
work_55utqx7tjrft5ojtbr67ypjdye	285	29	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	285	30	English	english	JJ
work_55utqx7tjrft5ojtbr67ypjdye	285	31	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	285	32	forms	form	NNS
work_55utqx7tjrft5ojtbr67ypjdye	285	33	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	286	1	Vocabulary	vocabulary	NN
work_55utqx7tjrft5ojtbr67ypjdye	286	2	curation	curation	NN
work_55utqx7tjrft5ojtbr67ypjdye	286	3	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	286	4	have	have	VB
work_55utqx7tjrft5ojtbr67ypjdye	286	5	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	286	6	profound	profound	JJ
work_55utqx7tjrft5ojtbr67ypjdye	286	7	effect	effect	NN
work_55utqx7tjrft5ojtbr67ypjdye	286	8	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	286	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	286	10	results	result	NNS
work_55utqx7tjrft5ojtbr67ypjdye	286	11	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	286	12	statistical	statistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	286	13	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	286	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	286	15	yet	yet	CC
work_55utqx7tjrft5ojtbr67ypjdye	286	16	procedures	procedure	NNS
work_55utqx7tjrft5ojtbr67ypjdye	286	17	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	286	18	vocabulary	vocabulary	JJ
work_55utqx7tjrft5ojtbr67ypjdye	286	19	curation	curation	NN
work_55utqx7tjrft5ojtbr67ypjdye	286	20	have	have	VBP
work_55utqx7tjrft5ojtbr67ypjdye	286	21	largely	largely	RB
work_55utqx7tjrft5ojtbr67ypjdye	286	22	been	be	VBN
work_55utqx7tjrft5ojtbr67ypjdye	286	23	left	leave	VBN
work_55utqx7tjrft5ojtbr67ypjdye	286	24	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	286	25	unex-	unex-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	286	26	amined	amine	VBD
work_55utqx7tjrft5ojtbr67ypjdye	286	27	convention	convention	NN
work_55utqx7tjrft5ojtbr67ypjdye	286	28	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	286	29	undocumented	undocumented	JJ
work_55utqx7tjrft5ojtbr67ypjdye	286	30	folk	folk	NN
work_55utqx7tjrft5ojtbr67ypjdye	286	31	wisdom	wisdom	NN
work_55utqx7tjrft5ojtbr67ypjdye	286	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	287	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	287	2	find	find	VBP
work_55utqx7tjrft5ojtbr67ypjdye	287	3	that	that	IN
work_55utqx7tjrft5ojtbr67ypjdye	287	4	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	287	5	commonly	commonly	RB
work_55utqx7tjrft5ojtbr67ypjdye	287	6	used	use	VBN
work_55utqx7tjrft5ojtbr67ypjdye	287	7	method	method	NN
work_55utqx7tjrft5ojtbr67ypjdye	287	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	287	9	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	287	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	287	11	provides	provide	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	287	12	little	little	JJ
work_55utqx7tjrft5ojtbr67ypjdye	287	13	measurable	measurable	JJ
work_55utqx7tjrft5ojtbr67ypjdye	287	14	benefit	benefit	NN
work_55utqx7tjrft5ojtbr67ypjdye	287	15	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	287	16	may	may	MD
work_55utqx7tjrft5ojtbr67ypjdye	287	17	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	287	18	fact	fact	NN
work_55utqx7tjrft5ojtbr67ypjdye	287	19	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	287	20	harmful	harmful	JJ
work_55utqx7tjrft5ojtbr67ypjdye	287	21	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	288	1	As	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	288	2	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	288	3	mining	mining	NN
work_55utqx7tjrft5ojtbr67ypjdye	288	4	becomes	become	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	288	5	more	more	RBR
work_55utqx7tjrft5ojtbr67ypjdye	288	6	influential	influential	JJ
work_55utqx7tjrft5ojtbr67ypjdye	288	7	outside	outside	JJ
work_55utqx7tjrft5ojtbr67ypjdye	288	8	core	core	NN
work_55utqx7tjrft5ojtbr67ypjdye	288	9	NLP	NLP	NNP
work_55utqx7tjrft5ojtbr67ypjdye	288	10	research	research	NN
work_55utqx7tjrft5ojtbr67ypjdye	288	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	288	12	more	more	JJR
work_55utqx7tjrft5ojtbr67ypjdye	288	13	attention	attention	NN
work_55utqx7tjrft5ojtbr67ypjdye	288	14	must	must	MD
work_55utqx7tjrft5ojtbr67ypjdye	288	15	be	be	VB
work_55utqx7tjrft5ojtbr67ypjdye	288	16	paid	pay	VBN
work_55utqx7tjrft5ojtbr67ypjdye	288	17	to	to	IN
work_55utqx7tjrft5ojtbr67ypjdye	288	18	these	these	DT
work_55utqx7tjrft5ojtbr67ypjdye	288	19	issues	issue	NNS
work_55utqx7tjrft5ojtbr67ypjdye	288	20	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	289	1	8	8	CD
work_55utqx7tjrft5ojtbr67ypjdye	289	2	Acknowledgements	acknowledgement	NNS
work_55utqx7tjrft5ojtbr67ypjdye	289	3	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	289	4	would	would	MD
work_55utqx7tjrft5ojtbr67ypjdye	289	5	like	like	VB
work_55utqx7tjrft5ojtbr67ypjdye	289	6	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	289	7	thank	thank	VB
work_55utqx7tjrft5ojtbr67ypjdye	289	8	Jacob	Jacob	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	9	Gardner	Gardner	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	289	11	Jack	Jack	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	12	Hessel	Hessel	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	289	14	Andrew	Andrew	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	15	Loeb	Loeb	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	289	17	Brian	Brian	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	18	McInnis	McInnis	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	19	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	289	20	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	289	21	Elly	elly	RB
work_55utqx7tjrft5ojtbr67ypjdye	289	22	Schofield	Schofield	NNP
work_55utqx7tjrft5ojtbr67ypjdye	289	23	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	289	24	helping	help	VBG
work_55utqx7tjrft5ojtbr67ypjdye	289	25	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	289	26	refine	refine	VB
work_55utqx7tjrft5ojtbr67ypjdye	289	27	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	289	28	writing	writing	NN
work_55utqx7tjrft5ojtbr67ypjdye	289	29	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	289	30	this	this	DT
work_55utqx7tjrft5ojtbr67ypjdye	289	31	paper	paper	NN
work_55utqx7tjrft5ojtbr67ypjdye	289	32	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	290	1	We	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	290	2	also	also	RB
work_55utqx7tjrft5ojtbr67ypjdye	290	3	would	would	MD
work_55utqx7tjrft5ojtbr67ypjdye	290	4	like	like	VB
work_55utqx7tjrft5ojtbr67ypjdye	290	5	to	to	TO
work_55utqx7tjrft5ojtbr67ypjdye	290	6	thank	thank	VB
work_55utqx7tjrft5ojtbr67ypjdye	290	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	290	8	TACL	TACL	NNP
work_55utqx7tjrft5ojtbr67ypjdye	290	9	editors	editor	NNS
work_55utqx7tjrft5ojtbr67ypjdye	290	10	Mark	Mark	NNP
work_55utqx7tjrft5ojtbr67ypjdye	290	11	John-	John-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	290	12	son	son	NN
work_55utqx7tjrft5ojtbr67ypjdye	290	13	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	290	14	Hal	Hal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	290	15	Daumé	Daumé	NNP
work_55utqx7tjrft5ojtbr67ypjdye	290	16	III	iii	CD
work_55utqx7tjrft5ojtbr67ypjdye	290	17	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	290	18	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	290	19	reviewers	reviewer	NNS
work_55utqx7tjrft5ojtbr67ypjdye	290	20	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	290	21	their	-PRON-	PRP$
work_55utqx7tjrft5ojtbr67ypjdye	290	22	thoughtful	thoughtful	JJ
work_55utqx7tjrft5ojtbr67ypjdye	290	23	comments	comment	NNS
work_55utqx7tjrft5ojtbr67ypjdye	290	24	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	290	25	suggestions	suggestion	NNS
work_55utqx7tjrft5ojtbr67ypjdye	290	26	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	291	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	291	2	first	first	JJ
work_55utqx7tjrft5ojtbr67ypjdye	291	3	au-	au-	NN
work_55utqx7tjrft5ojtbr67ypjdye	291	4	thor	thor	NN
work_55utqx7tjrft5ojtbr67ypjdye	291	5	was	be	VBD
work_55utqx7tjrft5ojtbr67ypjdye	291	6	funded	fund	VBN
work_55utqx7tjrft5ojtbr67ypjdye	291	7	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	291	8	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	291	9	Cornell	Cornell	NNP
work_55utqx7tjrft5ojtbr67ypjdye	291	10	University	University	NNP
work_55utqx7tjrft5ojtbr67ypjdye	291	11	Fellowship	Fellowship	NNP
work_55utqx7tjrft5ojtbr67ypjdye	291	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	292	1	298	298	CD
work_55utqx7tjrft5ojtbr67ypjdye	292	2	References	reference	NNS
work_55utqx7tjrft5ojtbr67ypjdye	292	3	Narayan	Narayan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	292	4	L	L	NNP
work_55utqx7tjrft5ojtbr67ypjdye	292	5	Bhamidipati	Bhamidipati	NNP
work_55utqx7tjrft5ojtbr67ypjdye	292	6	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	292	7	Sankar	Sankar	NNP
work_55utqx7tjrft5ojtbr67ypjdye	292	8	K	K	NNP
work_55utqx7tjrft5ojtbr67ypjdye	292	9	Pal	Pal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	292	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	293	1	2007	2007	CD
work_55utqx7tjrft5ojtbr67ypjdye	293	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	294	1	Stem-	Stem-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	294	2	ming	ming	NNP
work_55utqx7tjrft5ojtbr67ypjdye	294	3	via	via	IN
work_55utqx7tjrft5ojtbr67ypjdye	294	4	distribution	distribution	NN
work_55utqx7tjrft5ojtbr67ypjdye	294	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	294	6	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	294	7	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	294	8	segregation	segregation	NN
work_55utqx7tjrft5ojtbr67ypjdye	294	9	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	294	10	clas-	clas-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	294	11	sification	sification	NNP
work_55utqx7tjrft5ojtbr67ypjdye	294	12	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	294	13	retrieval	retrieval	NN
work_55utqx7tjrft5ojtbr67ypjdye	294	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	295	1	Systems	system	NNS
work_55utqx7tjrft5ojtbr67ypjdye	295	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	295	3	Man	Man	NNP
work_55utqx7tjrft5ojtbr67ypjdye	295	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	295	5	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	295	6	Cyber-	Cyber-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	295	7	netics	netic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	295	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	295	9	Part	Part	NNP
work_55utqx7tjrft5ojtbr67ypjdye	295	10	B	B	NNP
work_55utqx7tjrft5ojtbr67ypjdye	295	11	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	295	12	Cybernetics	cybernetic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	295	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	295	14	IEEE	IEEE	NNP
work_55utqx7tjrft5ojtbr67ypjdye	295	15	Transactions	transaction	NNS
work_55utqx7tjrft5ojtbr67ypjdye	295	16	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	295	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	295	18	37(2):350–360	37(2):350–360	CD
work_55utqx7tjrft5ojtbr67ypjdye	295	19	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	296	1	Steven	Steven	NNP
work_55utqx7tjrft5ojtbr67ypjdye	296	2	Bird	Bird	NNP
work_55utqx7tjrft5ojtbr67ypjdye	296	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	296	4	Ewan	Ewan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	296	5	Klein	Klein	NNP
work_55utqx7tjrft5ojtbr67ypjdye	296	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	296	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	296	8	Edward	Edward	NNP
work_55utqx7tjrft5ojtbr67ypjdye	296	9	Loper	Loper	NNP
work_55utqx7tjrft5ojtbr67ypjdye	296	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	297	1	2009	2009	CD
work_55utqx7tjrft5ojtbr67ypjdye	297	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	298	1	Nat-	nat-	DT
work_55utqx7tjrft5ojtbr67ypjdye	298	2	ural	ural	JJ
work_55utqx7tjrft5ojtbr67ypjdye	298	3	Language	Language	NNP
work_55utqx7tjrft5ojtbr67ypjdye	298	4	Processing	Processing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	298	5	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	298	6	Python	Python	NNP
work_55utqx7tjrft5ojtbr67ypjdye	298	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	299	1	O’Reilly	o’reilly	RB
work_55utqx7tjrft5ojtbr67ypjdye	299	2	Me-	me-	CD
work_55utqx7tjrft5ojtbr67ypjdye	299	3	dia	dia	NN
work_55utqx7tjrft5ojtbr67ypjdye	299	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	300	1	Available	available	JJ
work_55utqx7tjrft5ojtbr67ypjdye	300	2	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	300	3	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	300	4	http://www.nltk.org/book/.	http://www.nltk.org/book/.	ADD
work_55utqx7tjrft5ojtbr67ypjdye	301	1	David	David	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	2	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	3	Blei	Blei	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	301	5	Andrew	Andrew	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	6	Y	Y	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	7	Ng	Ng	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	301	9	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	301	10	Michael	Michael	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	11	I	I	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	12	Jordan	Jordan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	301	13	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	302	1	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	302	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	303	1	Latent	latent	JJ
work_55utqx7tjrft5ojtbr67ypjdye	303	2	dirichlet	dirichlet	NN
work_55utqx7tjrft5ojtbr67ypjdye	303	3	allocation	allocation	NN
work_55utqx7tjrft5ojtbr67ypjdye	303	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	304	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	304	2	Journal	Journal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	304	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	304	4	Ma-	Ma-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	304	5	chine	chine	NN
work_55utqx7tjrft5ojtbr67ypjdye	304	6	Learning	Learning	NNP
work_55utqx7tjrft5ojtbr67ypjdye	304	7	Research	Research	NNP
work_55utqx7tjrft5ojtbr67ypjdye	304	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	304	9	3:993–1022	3:993–1022	CD
work_55utqx7tjrft5ojtbr67ypjdye	304	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	305	1	Matt	Matt	NNP
work_55utqx7tjrft5ojtbr67ypjdye	305	2	Chaput	Chaput	NNP
work_55utqx7tjrft5ojtbr67ypjdye	305	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	306	1	2010	2010	CD
work_55utqx7tjrft5ojtbr67ypjdye	306	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	307	1	Stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	307	2	library	library	NN
work_55utqx7tjrft5ojtbr67ypjdye	307	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	308	1	Available	available	JJ
work_55utqx7tjrft5ojtbr67ypjdye	308	2	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	308	3	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	308	4	https://bitbucket.org/mchaput/stemming	https://bitbucket.org/mchaput/stemming	NN
work_55utqx7tjrft5ojtbr67ypjdye	308	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	309	1	Antske	Antske	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	2	Fokkens	Fokkens	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	309	4	Marieke	Marieke	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	5	Van	Van	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	6	Erp	Erp	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	309	8	Marten	Marten	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	9	Postma	Postma	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	309	11	Ted	Ted	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	12	Pedersen	Pedersen	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	309	14	Piek	Piek	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	15	Vossen	Vossen	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	309	17	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	309	18	Nuno	Nuno	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	19	Freire	Freire	NNP
work_55utqx7tjrft5ojtbr67ypjdye	309	20	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	310	1	2013	2013	CD
work_55utqx7tjrft5ojtbr67ypjdye	310	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	311	1	Off-	off-	DT
work_55utqx7tjrft5ojtbr67ypjdye	311	2	spring	spring	NN
work_55utqx7tjrft5ojtbr67ypjdye	311	3	from	from	IN
work_55utqx7tjrft5ojtbr67ypjdye	311	4	reproduction	reproduction	NN
work_55utqx7tjrft5ojtbr67ypjdye	311	5	problems	problem	NNS
work_55utqx7tjrft5ojtbr67ypjdye	311	6	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	311	7	What	what	WDT
work_55utqx7tjrft5ojtbr67ypjdye	311	8	replication	replication	NN
work_55utqx7tjrft5ojtbr67ypjdye	311	9	failure	failure	NN
work_55utqx7tjrft5ojtbr67ypjdye	311	10	teaches	teach	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	311	11	us	-PRON-	PRP
work_55utqx7tjrft5ojtbr67ypjdye	311	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	312	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	312	2	Proceedings	proceeding	NNS
work_55utqx7tjrft5ojtbr67ypjdye	312	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	312	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	312	5	51st	51st	JJ
work_55utqx7tjrft5ojtbr67ypjdye	312	6	ACL	ACL	NNP
work_55utqx7tjrft5ojtbr67ypjdye	312	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	312	8	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	312	9	1691–1701	1691–1701	CD
work_55utqx7tjrft5ojtbr67ypjdye	312	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	313	1	William	William	NNP
work_55utqx7tjrft5ojtbr67ypjdye	313	2	B	B	NNP
work_55utqx7tjrft5ojtbr67ypjdye	313	3	Frakes	Frakes	NNP
work_55utqx7tjrft5ojtbr67ypjdye	313	4	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	313	5	Christopher	Christopher	NNP
work_55utqx7tjrft5ojtbr67ypjdye	313	6	J	J	NNP
work_55utqx7tjrft5ojtbr67ypjdye	313	7	Fox	Fox	NNP
work_55utqx7tjrft5ojtbr67ypjdye	313	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	314	1	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	314	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	315	1	Strength	strength	NN
work_55utqx7tjrft5ojtbr67ypjdye	315	2	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	315	3	similarity	similarity	NN
work_55utqx7tjrft5ojtbr67ypjdye	315	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	315	5	affix	affix	NNP
work_55utqx7tjrft5ojtbr67ypjdye	315	6	removal	removal	NN
work_55utqx7tjrft5ojtbr67ypjdye	315	7	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	315	8	algorithms	algorithm	NNS
work_55utqx7tjrft5ojtbr67ypjdye	315	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	316	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	316	2	ACM	ACM	NNP
work_55utqx7tjrft5ojtbr67ypjdye	316	3	SIGIR	SIGIR	NNP
work_55utqx7tjrft5ojtbr67ypjdye	316	4	Forum	Forum	NNP
work_55utqx7tjrft5ojtbr67ypjdye	316	5	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	316	6	volume	volume	NN
work_55utqx7tjrft5ojtbr67ypjdye	316	7	37	37	CD
work_55utqx7tjrft5ojtbr67ypjdye	316	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	316	9	pages	page	NNS
work_55utqx7tjrft5ojtbr67ypjdye	316	10	26–30	26–30	VBP
work_55utqx7tjrft5ojtbr67ypjdye	316	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	317	1	ACM	ACM	NNP
work_55utqx7tjrft5ojtbr67ypjdye	317	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	318	1	Kuzman	Kuzman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	318	2	Ganchev	Ganchev	NNP
work_55utqx7tjrft5ojtbr67ypjdye	318	3	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	318	4	Mark	Mark	NNP
work_55utqx7tjrft5ojtbr67ypjdye	318	5	Dredze	Dredze	NNP
work_55utqx7tjrft5ojtbr67ypjdye	318	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	319	1	2008	2008	CD
work_55utqx7tjrft5ojtbr67ypjdye	319	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	320	1	Small	small	JJ
work_55utqx7tjrft5ojtbr67ypjdye	320	2	sta-	sta-	NN
work_55utqx7tjrft5ojtbr67ypjdye	320	3	tistical	tistical	JJ
work_55utqx7tjrft5ojtbr67ypjdye	320	4	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	320	5	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	320	6	random	random	JJ
work_55utqx7tjrft5ojtbr67ypjdye	320	7	feature	feature	NN
work_55utqx7tjrft5ojtbr67ypjdye	320	8	mixing	mix	VBG
work_55utqx7tjrft5ojtbr67ypjdye	320	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	321	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	321	2	Proceed-	Proceed-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	321	3	ings	ing	NNS
work_55utqx7tjrft5ojtbr67ypjdye	321	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	321	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	321	6	ACL08	acl08	JJ
work_55utqx7tjrft5ojtbr67ypjdye	321	7	HLT	HLT	NNP
work_55utqx7tjrft5ojtbr67ypjdye	321	8	Workshop	Workshop	NNP
work_55utqx7tjrft5ojtbr67ypjdye	321	9	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	321	10	Mobile	Mobile	NNP
work_55utqx7tjrft5ojtbr67ypjdye	321	11	Language	Language	NNP
work_55utqx7tjrft5ojtbr67ypjdye	321	12	Processing	Processing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	321	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	321	14	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	321	15	19–20	19–20	CD
work_55utqx7tjrft5ojtbr67ypjdye	321	16	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	322	1	Justin	Justin	NNP
work_55utqx7tjrft5ojtbr67ypjdye	322	2	Grimmer	Grimmer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	322	3	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	322	4	Gary	Gary	NNP
work_55utqx7tjrft5ojtbr67ypjdye	322	5	King	King	NNP
work_55utqx7tjrft5ojtbr67ypjdye	322	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	323	1	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	323	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	324	1	General	general	JJ
work_55utqx7tjrft5ojtbr67ypjdye	324	2	pur-	pur-	DT
work_55utqx7tjrft5ojtbr67ypjdye	324	3	pose	pose	VBP
work_55utqx7tjrft5ojtbr67ypjdye	324	4	computer	computer	NN
work_55utqx7tjrft5ojtbr67ypjdye	324	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	324	6	assisted	assist	VBN
work_55utqx7tjrft5ojtbr67ypjdye	324	7	clustering	clustering	NN
work_55utqx7tjrft5ojtbr67ypjdye	324	8	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	324	9	conceptualiza-	conceptualiza-	NN
work_55utqx7tjrft5ojtbr67ypjdye	324	10	tion	tion	NN
work_55utqx7tjrft5ojtbr67ypjdye	324	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	325	1	PNAS	pna	NNS
work_55utqx7tjrft5ojtbr67ypjdye	325	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	325	3	108(7):2643–2650	108(7):2643–2650	CD
work_55utqx7tjrft5ojtbr67ypjdye	325	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	326	1	Pu	Pu	NNP
work_55utqx7tjrft5ojtbr67ypjdye	326	2	Han	Han	NNP
work_55utqx7tjrft5ojtbr67ypjdye	326	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	326	4	Si	Si	NNP
work_55utqx7tjrft5ojtbr67ypjdye	326	5	Shen	Shen	NNP
work_55utqx7tjrft5ojtbr67ypjdye	326	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	326	7	Dongbo	Dongbo	NNP
work_55utqx7tjrft5ojtbr67ypjdye	326	8	Wang	Wang	NNP
work_55utqx7tjrft5ojtbr67ypjdye	326	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	326	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	326	11	Yanyun	Yanyun	NNP
work_55utqx7tjrft5ojtbr67ypjdye	326	12	Liu	Liu	NNP
work_55utqx7tjrft5ojtbr67ypjdye	326	13	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	327	1	2012	2012	CD
work_55utqx7tjrft5ojtbr67ypjdye	327	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	328	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	328	2	influence	influence	NN
work_55utqx7tjrft5ojtbr67ypjdye	328	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	328	4	word	word	NN
work_55utqx7tjrft5ojtbr67ypjdye	328	5	normalization	normalization	NN
work_55utqx7tjrft5ojtbr67ypjdye	328	6	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	328	7	english	english	NNP
work_55utqx7tjrft5ojtbr67ypjdye	328	8	docu-	docu-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	328	9	ment	ment	JJ
work_55utqx7tjrft5ojtbr67ypjdye	328	10	clustering	clustering	NN
work_55utqx7tjrft5ojtbr67ypjdye	328	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	329	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	329	2	Computer	Computer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	329	3	Science	Science	NNP
work_55utqx7tjrft5ojtbr67ypjdye	329	4	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	329	5	Automation	Automation	NNP
work_55utqx7tjrft5ojtbr67ypjdye	329	6	Engineering	Engineering	NNP
work_55utqx7tjrft5ojtbr67ypjdye	329	7	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	329	8	CSAE	CSAE	NNP
work_55utqx7tjrft5ojtbr67ypjdye	329	9	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	329	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	329	11	2012	2012	CD
work_55utqx7tjrft5ojtbr67ypjdye	329	12	IEEE	IEEE	NNP
work_55utqx7tjrft5ojtbr67ypjdye	329	13	International	International	NNP
work_55utqx7tjrft5ojtbr67ypjdye	329	14	Con-	Con-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	329	15	ference	ference	NN
work_55utqx7tjrft5ojtbr67ypjdye	329	16	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	329	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	329	18	volume	volume	NN
work_55utqx7tjrft5ojtbr67ypjdye	329	19	2	2	CD
work_55utqx7tjrft5ojtbr67ypjdye	329	20	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	329	21	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	329	22	116–120	116–120	CD
work_55utqx7tjrft5ojtbr67ypjdye	329	23	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	330	1	IEEE	IEEE	NNP
work_55utqx7tjrft5ojtbr67ypjdye	330	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	331	1	Donna	Donna	NNP
work_55utqx7tjrft5ojtbr67ypjdye	331	2	Harman	Harman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	331	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	332	1	1991	1991	CD
work_55utqx7tjrft5ojtbr67ypjdye	332	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	333	1	How	how	WRB
work_55utqx7tjrft5ojtbr67ypjdye	333	2	effective	effective	JJ
work_55utqx7tjrft5ojtbr67ypjdye	333	3	is	be	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	333	4	suffixing	suffix	VBG
work_55utqx7tjrft5ojtbr67ypjdye	333	5	?	?	.
work_55utqx7tjrft5ojtbr67ypjdye	334	1	Jour-	Jour-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	334	2	nal	nal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	334	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	334	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	334	5	American	American	NNP
work_55utqx7tjrft5ojtbr67ypjdye	334	6	Society	Society	NNP
work_55utqx7tjrft5ojtbr67ypjdye	334	7	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	334	8	Information	Information	NNP
work_55utqx7tjrft5ojtbr67ypjdye	334	9	Science	Science	NNP
work_55utqx7tjrft5ojtbr67ypjdye	334	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	334	11	42(1):7–15	42(1):7–15	CD
work_55utqx7tjrft5ojtbr67ypjdye	334	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	335	1	Carina	Carina	NNP
work_55utqx7tjrft5ojtbr67ypjdye	335	2	Jacobi	Jacobi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	335	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	335	4	Wouter	Wouter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	335	5	van	van	NNP
work_55utqx7tjrft5ojtbr67ypjdye	335	6	Atteveldt	Atteveldt	NNP
work_55utqx7tjrft5ojtbr67ypjdye	335	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	335	8	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	335	9	Kasper	Kasper	NNP
work_55utqx7tjrft5ojtbr67ypjdye	335	10	Wel-	Wel-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	335	11	bers	ber	NNS
work_55utqx7tjrft5ojtbr67ypjdye	335	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	336	1	2016	2016	CD
work_55utqx7tjrft5ojtbr67ypjdye	336	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	337	1	Quantitative	quantitative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	337	2	analysis	analysis	NN
work_55utqx7tjrft5ojtbr67ypjdye	337	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	337	4	large	large	JJ
work_55utqx7tjrft5ojtbr67ypjdye	337	5	amounts	amount	NNS
work_55utqx7tjrft5ojtbr67ypjdye	337	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	337	7	journalistic	journalistic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	337	8	texts	text	NNS
work_55utqx7tjrft5ojtbr67ypjdye	337	9	using	use	VBG
work_55utqx7tjrft5ojtbr67ypjdye	337	10	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	337	11	modelling	modelling	NN
work_55utqx7tjrft5ojtbr67ypjdye	337	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	338	1	Digital	Digital	NNP
work_55utqx7tjrft5ojtbr67ypjdye	338	2	Jour-	Jour-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	338	3	nalism	nalism	NN
work_55utqx7tjrft5ojtbr67ypjdye	338	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	338	5	4(1):89–106	4(1):89–106	CD
work_55utqx7tjrft5ojtbr67ypjdye	338	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	339	1	Anjali	Anjali	NNP
work_55utqx7tjrft5ojtbr67ypjdye	339	2	Ganesh	Ganesh	NNP
work_55utqx7tjrft5ojtbr67ypjdye	339	3	Jivani	Jivani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	339	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	340	1	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	340	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	341	1	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	341	2	comparative	comparative	JJ
work_55utqx7tjrft5ojtbr67ypjdye	341	3	study	study	NN
work_55utqx7tjrft5ojtbr67ypjdye	341	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	341	5	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	341	6	algorithms	algorithm	NNS
work_55utqx7tjrft5ojtbr67ypjdye	341	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	342	1	International	International	NNP
work_55utqx7tjrft5ojtbr67ypjdye	342	2	Journal	Journal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	342	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	342	4	Com-	Com-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	342	5	puter	puter	NN
work_55utqx7tjrft5ojtbr67ypjdye	342	6	Technology	Technology	NNP
work_55utqx7tjrft5ojtbr67ypjdye	342	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	342	8	Applications	Applications	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	342	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	342	10	2(6):1930–1938	2(6):1930–1938	CD
work_55utqx7tjrft5ojtbr67ypjdye	342	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	343	1	Matthew	Matthew	NNP
work_55utqx7tjrft5ojtbr67ypjdye	343	2	L	L	NNP
work_55utqx7tjrft5ojtbr67ypjdye	343	3	Jockers	Jockers	NNP
work_55utqx7tjrft5ojtbr67ypjdye	343	4	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	343	5	David	David	NNP
work_55utqx7tjrft5ojtbr67ypjdye	343	6	Mimno	Mimno	NNP
work_55utqx7tjrft5ojtbr67ypjdye	343	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	344	1	2013	2013	CD
work_55utqx7tjrft5ojtbr67ypjdye	344	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	345	1	Significant	significant	JJ
work_55utqx7tjrft5ojtbr67ypjdye	345	2	themes	theme	NNS
work_55utqx7tjrft5ojtbr67ypjdye	345	3	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	345	4	19th	19th	JJ
work_55utqx7tjrft5ojtbr67ypjdye	345	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	345	6	century	century	NN
work_55utqx7tjrft5ojtbr67ypjdye	345	7	literature	literature	NN
work_55utqx7tjrft5ojtbr67ypjdye	345	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	346	1	Poetics	poetic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	346	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	346	3	41(6):750	41(6):750	CD
work_55utqx7tjrft5ojtbr67ypjdye	346	4	–	–	:
work_55utqx7tjrft5ojtbr67ypjdye	346	5	769	769	CD
work_55utqx7tjrft5ojtbr67ypjdye	346	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	347	1	Sowmya	Sowmya	NNP
work_55utqx7tjrft5ojtbr67ypjdye	347	2	Kamath	Kamath	NNP
work_55utqx7tjrft5ojtbr67ypjdye	347	3	S	S	NNP
work_55utqx7tjrft5ojtbr67ypjdye	347	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	347	5	Atif	Atif	NNP
work_55utqx7tjrft5ojtbr67ypjdye	347	6	Ahmed	Ahmed	NNP
work_55utqx7tjrft5ojtbr67ypjdye	347	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	347	8	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	347	9	Mani	Mani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	347	10	Shankar	Shankar	NNP
work_55utqx7tjrft5ojtbr67ypjdye	347	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	348	1	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	348	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	349	1	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	349	2	composite	composite	JJ
work_55utqx7tjrft5ojtbr67ypjdye	349	3	classification	classification	NN
work_55utqx7tjrft5ojtbr67ypjdye	349	4	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	349	5	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	349	6	web	web	NN
work_55utqx7tjrft5ojtbr67ypjdye	349	7	ser-	ser-	NN
work_55utqx7tjrft5ojtbr67ypjdye	349	8	vices	vice	NNS
work_55utqx7tjrft5ojtbr67ypjdye	349	9	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	349	10	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	349	11	semantic	semantic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	349	12	&	&	CC
work_55utqx7tjrft5ojtbr67ypjdye	349	13	syntactic	syntactic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	349	14	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	349	15	inte-	inte-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	349	16	gration	gration	NN
work_55utqx7tjrft5ojtbr67ypjdye	349	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	350	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	350	2	Advance	Advance	NNP
work_55utqx7tjrft5ojtbr67ypjdye	350	3	Computing	Computing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	350	4	Conference	Conference	NNP
work_55utqx7tjrft5ojtbr67ypjdye	350	5	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	350	6	IACC	IACC	NNP
work_55utqx7tjrft5ojtbr67ypjdye	350	7	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	350	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	350	9	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	350	10	IEEE	IEEE	NNP
work_55utqx7tjrft5ojtbr67ypjdye	350	11	International	International	NNP
work_55utqx7tjrft5ojtbr67ypjdye	350	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	350	13	pages	page	NNS
work_55utqx7tjrft5ojtbr67ypjdye	350	14	1169–1173	1169–1173	CD
work_55utqx7tjrft5ojtbr67ypjdye	350	15	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	351	1	IEEE	IEEE	NNP
work_55utqx7tjrft5ojtbr67ypjdye	351	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	352	1	Robert	Robert	NNP
work_55utqx7tjrft5ojtbr67ypjdye	352	2	Krovetz	Krovetz	NNP
work_55utqx7tjrft5ojtbr67ypjdye	352	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	353	1	1993	1993	CD
work_55utqx7tjrft5ojtbr67ypjdye	353	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	354	1	Viewing	view	VBG
work_55utqx7tjrft5ojtbr67ypjdye	354	2	morphology	morphology	NN
work_55utqx7tjrft5ojtbr67ypjdye	354	3	as	as	IN
work_55utqx7tjrft5ojtbr67ypjdye	354	4	an	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	354	5	in-	in-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	354	6	ference	ference	NN
work_55utqx7tjrft5ojtbr67ypjdye	354	7	process	process	NN
work_55utqx7tjrft5ojtbr67ypjdye	354	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	355	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	355	2	Proceedings	proceeding	NNS
work_55utqx7tjrft5ojtbr67ypjdye	355	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	355	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	355	5	16th	16th	JJ
work_55utqx7tjrft5ojtbr67ypjdye	355	6	annual	annual	JJ
work_55utqx7tjrft5ojtbr67ypjdye	355	7	international	international	JJ
work_55utqx7tjrft5ojtbr67ypjdye	355	8	ACM	ACM	NNP
work_55utqx7tjrft5ojtbr67ypjdye	355	9	SIGIR	SIGIR	NNP
work_55utqx7tjrft5ojtbr67ypjdye	355	10	conference	conference	NN
work_55utqx7tjrft5ojtbr67ypjdye	355	11	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	355	12	Research	research	NN
work_55utqx7tjrft5ojtbr67ypjdye	355	13	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	355	14	development	development	NN
work_55utqx7tjrft5ojtbr67ypjdye	355	15	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	355	16	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	355	17	retrieval	retrieval	NN
work_55utqx7tjrft5ojtbr67ypjdye	355	18	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	355	19	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	355	20	191–202	191–202	CD
work_55utqx7tjrft5ojtbr67ypjdye	355	21	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	356	1	ACM	ACM	NNP
work_55utqx7tjrft5ojtbr67ypjdye	356	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	357	1	Jey	Jey	NNP
work_55utqx7tjrft5ojtbr67ypjdye	357	2	Han	Han	NNP
work_55utqx7tjrft5ojtbr67ypjdye	357	3	Lau	Lau	NNP
work_55utqx7tjrft5ojtbr67ypjdye	357	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	357	5	David	David	NNP
work_55utqx7tjrft5ojtbr67ypjdye	357	6	Newman	Newman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	357	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	357	8	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	357	9	Timothy	Timothy	NNP
work_55utqx7tjrft5ojtbr67ypjdye	357	10	Baldwin	Baldwin	NNP
work_55utqx7tjrft5ojtbr67ypjdye	357	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	358	1	2014	2014	CD
work_55utqx7tjrft5ojtbr67ypjdye	358	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	359	1	Machine	machine	NN
work_55utqx7tjrft5ojtbr67ypjdye	359	2	reading	read	VBG
work_55utqx7tjrft5ojtbr67ypjdye	359	3	tea	tea	NN
work_55utqx7tjrft5ojtbr67ypjdye	359	4	leaves	leave	NNS
work_55utqx7tjrft5ojtbr67ypjdye	359	5	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	359	6	Automatically	automatically	RB
work_55utqx7tjrft5ojtbr67ypjdye	359	7	evaluating	evaluate	VBG
work_55utqx7tjrft5ojtbr67ypjdye	359	8	topic	topic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	359	9	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	359	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	359	11	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	359	12	model	model	NN
work_55utqx7tjrft5ojtbr67ypjdye	359	13	quality	quality	NN
work_55utqx7tjrft5ojtbr67ypjdye	359	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	360	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	360	2	Proceedings	Proceedings	NNP
work_55utqx7tjrft5ojtbr67ypjdye	360	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	360	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	360	5	Association	Association	NNP
work_55utqx7tjrft5ojtbr67ypjdye	360	6	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	360	7	Computational	Computational	NNP
work_55utqx7tjrft5ojtbr67ypjdye	360	8	Lin-	Lin-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	360	9	guistics	guistic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	360	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	360	11	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	360	12	530–539	530–539	CD
work_55utqx7tjrft5ojtbr67ypjdye	360	13	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	361	1	Zhiyuan	Zhiyuan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	361	2	Liu	Liu	NNP
work_55utqx7tjrft5ojtbr67ypjdye	361	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	361	4	Wenyi	Wenyi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	361	5	Huang	Huang	NNP
work_55utqx7tjrft5ojtbr67ypjdye	361	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	361	7	Yabin	Yabin	NNP
work_55utqx7tjrft5ojtbr67ypjdye	361	8	Zheng	Zheng	NNP
work_55utqx7tjrft5ojtbr67ypjdye	361	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	361	10	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	361	11	Maosong	Maosong	NNP
work_55utqx7tjrft5ojtbr67ypjdye	361	12	Sun	Sun	NNP
work_55utqx7tjrft5ojtbr67ypjdye	361	13	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	362	1	2010	2010	CD
work_55utqx7tjrft5ojtbr67ypjdye	362	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	363	1	Automatic	automatic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	363	2	keyphrase	keyphrase	NN
work_55utqx7tjrft5ojtbr67ypjdye	363	3	extraction	extraction	NN
work_55utqx7tjrft5ojtbr67ypjdye	363	4	via	via	IN
work_55utqx7tjrft5ojtbr67ypjdye	363	5	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	363	6	decomposition	decomposition	NN
work_55utqx7tjrft5ojtbr67ypjdye	363	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	364	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	364	2	Proceedings	Proceedings	NNP
work_55utqx7tjrft5ojtbr67ypjdye	364	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	364	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	364	5	Conference	Conference	NNP
work_55utqx7tjrft5ojtbr67ypjdye	364	6	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	364	7	Empirical	Empirical	NNP
work_55utqx7tjrft5ojtbr67ypjdye	364	8	Methods	Methods	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	364	9	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	364	10	Natural	Natural	NNP
work_55utqx7tjrft5ojtbr67ypjdye	364	11	Language	Language	NNP
work_55utqx7tjrft5ojtbr67ypjdye	364	12	Processing	Processing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	364	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	364	14	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	364	15	366–376	366–376	CD
work_55utqx7tjrft5ojtbr67ypjdye	364	16	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	365	1	Association	Association	NNP
work_55utqx7tjrft5ojtbr67ypjdye	365	2	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	365	3	Computational	Computational	NNP
work_55utqx7tjrft5ojtbr67ypjdye	365	4	Lin-	Lin-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	365	5	guistics	guistic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	365	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	366	1	Siaw	Siaw	NNP
work_55utqx7tjrft5ojtbr67ypjdye	366	2	Ling	Ling	NNP
work_55utqx7tjrft5ojtbr67ypjdye	366	3	Lo	Lo	NNP
work_55utqx7tjrft5ojtbr67ypjdye	366	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	366	5	David	David	NNP
work_55utqx7tjrft5ojtbr67ypjdye	366	6	Cornforth	Cornforth	NNP
work_55utqx7tjrft5ojtbr67ypjdye	366	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	366	8	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	366	9	Raymond	Raymond	NNP
work_55utqx7tjrft5ojtbr67ypjdye	366	10	Chiong	Chiong	NNP
work_55utqx7tjrft5ojtbr67ypjdye	366	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	367	1	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	367	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	368	1	Effects	effect	NNS
work_55utqx7tjrft5ojtbr67ypjdye	368	2	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	368	3	training	training	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	4	datasets	dataset	NNS
work_55utqx7tjrft5ojtbr67ypjdye	368	5	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	368	6	both	both	CC
work_55utqx7tjrft5ojtbr67ypjdye	368	7	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	368	8	extreme	extreme	JJ
work_55utqx7tjrft5ojtbr67ypjdye	368	9	learning	learn	VBG
work_55utqx7tjrft5ojtbr67ypjdye	368	10	machine	machine	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	368	12	support	support	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	13	vector	vector	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	14	machine	machine	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	15	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	368	16	tar-	tar-	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	17	get	get	VBP
work_55utqx7tjrft5ojtbr67ypjdye	368	18	audience	audience	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	19	identification	identification	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	20	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	368	21	twitter	twitter	NN
work_55utqx7tjrft5ojtbr67ypjdye	368	22	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	369	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	369	2	Proceedings	Proceedings	NNP
work_55utqx7tjrft5ojtbr67ypjdye	369	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	369	4	ELM-2014	ELM-2014	NNP
work_55utqx7tjrft5ojtbr67ypjdye	369	5	Volume	volume	NN
work_55utqx7tjrft5ojtbr67ypjdye	369	6	1	1	CD
work_55utqx7tjrft5ojtbr67ypjdye	369	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	369	8	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	369	9	417–434	417–434	CD
work_55utqx7tjrft5ojtbr67ypjdye	369	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	370	1	Springer	Springer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	370	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	371	1	Julie	Julie	NNP
work_55utqx7tjrft5ojtbr67ypjdye	371	2	B	B	NNP
work_55utqx7tjrft5ojtbr67ypjdye	371	3	Lovins	Lovins	NNP
work_55utqx7tjrft5ojtbr67ypjdye	371	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	372	1	1968	1968	CD
work_55utqx7tjrft5ojtbr67ypjdye	372	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	373	1	Development	development	NN
work_55utqx7tjrft5ojtbr67ypjdye	373	2	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	373	3	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	373	4	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	373	5	al-	al-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	373	6	gorithm	gorithm	NN
work_55utqx7tjrft5ojtbr67ypjdye	373	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	374	1	Mechanical	Mechanical	NNP
work_55utqx7tjrft5ojtbr67ypjdye	374	2	Translation	Translation	NNP
work_55utqx7tjrft5ojtbr67ypjdye	374	3	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	374	4	Computational	Computational	NNP
work_55utqx7tjrft5ojtbr67ypjdye	374	5	Linguistics	Linguistics	NNP
work_55utqx7tjrft5ojtbr67ypjdye	374	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	374	7	11:22–31	11:22–31	CD
work_55utqx7tjrft5ojtbr67ypjdye	374	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	375	1	Prasenjit	Prasenjit	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	2	Majumder	Majumder	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	375	4	Mandar	Mandar	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	5	Mitra	Mitra	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	375	7	Swapan	Swapan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	8	K	K	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	9	Parui	Parui	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	375	11	Gobinda	Gobinda	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	12	Kole	Kole	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	375	14	Pabitra	Pabitra	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	15	Mitra	Mitra	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	16	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	375	17	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	375	18	Kalyankumar	Kalyankumar	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	19	Datta	Datta	NNP
work_55utqx7tjrft5ojtbr67ypjdye	375	20	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	376	1	2007	2007	CD
work_55utqx7tjrft5ojtbr67ypjdye	376	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	377	1	Yass	yass	NN
work_55utqx7tjrft5ojtbr67ypjdye	377	2	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	377	3	Yet	yet	RB
work_55utqx7tjrft5ojtbr67ypjdye	377	4	another	another	DT
work_55utqx7tjrft5ojtbr67ypjdye	377	5	suffix	suffix	NN
work_55utqx7tjrft5ojtbr67ypjdye	377	6	stripper	stripper	NN
work_55utqx7tjrft5ojtbr67ypjdye	377	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	378	1	ACM	ACM	NNP
work_55utqx7tjrft5ojtbr67ypjdye	378	2	Trans-	Trans-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	378	3	actions	action	NNS
work_55utqx7tjrft5ojtbr67ypjdye	378	4	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	378	5	Information	Information	NNP
work_55utqx7tjrft5ojtbr67ypjdye	378	6	Systems	Systems	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	378	7	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	378	8	TOIS	TOIS	NNP
work_55utqx7tjrft5ojtbr67ypjdye	378	9	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	378	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	378	11	25(4):18	25(4):18	NNP
work_55utqx7tjrft5ojtbr67ypjdye	378	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	379	1	Andrew	Andrew	NNP
work_55utqx7tjrft5ojtbr67ypjdye	379	2	K	K	NNP
work_55utqx7tjrft5ojtbr67ypjdye	379	3	McCallum	McCallum	NNP
work_55utqx7tjrft5ojtbr67ypjdye	379	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	380	1	2002	2002	CD
work_55utqx7tjrft5ojtbr67ypjdye	380	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	381	1	Mallet	mallet	NN
work_55utqx7tjrft5ojtbr67ypjdye	381	2	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	381	3	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	381	4	ma-	ma-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	381	5	chine	chine	NN
work_55utqx7tjrft5ojtbr67ypjdye	381	6	learning	learn	VBG
work_55utqx7tjrft5ojtbr67ypjdye	381	7	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	381	8	language	language	NN
work_55utqx7tjrft5ojtbr67ypjdye	381	9	toolkit	toolkit	NNS
work_55utqx7tjrft5ojtbr67ypjdye	381	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	382	1	Available	available	JJ
work_55utqx7tjrft5ojtbr67ypjdye	382	2	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	382	3	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	382	4	http://mallet.cs.umass.edu	http://mallet.cs.umass.edu	NNS
work_55utqx7tjrft5ojtbr67ypjdye	382	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	383	1	Marina	Marina	NNP
work_55utqx7tjrft5ojtbr67ypjdye	383	2	Meilă.	Meilă.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	384	1	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	384	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	385	1	Comparing	compare	VBG
work_55utqx7tjrft5ojtbr67ypjdye	385	2	clusterings	clustering	NNS
work_55utqx7tjrft5ojtbr67ypjdye	385	3	by	by	IN
work_55utqx7tjrft5ojtbr67ypjdye	385	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	385	5	vari-	vari-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	385	6	ation	ation	NN
work_55utqx7tjrft5ojtbr67ypjdye	385	7	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	385	8	information	information	NN
work_55utqx7tjrft5ojtbr67ypjdye	385	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	386	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	386	2	Bernhard	Bernhard	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	3	Schölkopf	Schölkopf	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	4	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	386	5	Man-	Man-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	6	fred	fred	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	7	K.	K.	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	8	Warmuth	Warmuth	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	386	10	editors	editor	NNS
work_55utqx7tjrft5ojtbr67ypjdye	386	11	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	386	12	Learning	Learning	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	13	Theory	Theory	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	14	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	386	15	Kernel	Kernel	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	16	Machines	Machines	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	386	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	386	18	volume	volume	NN
work_55utqx7tjrft5ojtbr67ypjdye	386	19	2777	2777	CD
work_55utqx7tjrft5ojtbr67ypjdye	386	20	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	386	21	Lecture	Lecture	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	22	Notes	Notes	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	386	23	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	386	24	Computer	Computer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	25	Science	Science	NNP
work_55utqx7tjrft5ojtbr67ypjdye	386	26	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	386	27	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	386	28	173–187	173–187	CD
work_55utqx7tjrft5ojtbr67ypjdye	386	29	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	387	1	Springer	Springer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	387	2	Berlin	Berlin	NNP
work_55utqx7tjrft5ojtbr67ypjdye	387	3	Heidelberg	Heidelberg	NNP
work_55utqx7tjrft5ojtbr67ypjdye	387	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	388	1	Massimo	Massimo	NNP
work_55utqx7tjrft5ojtbr67ypjdye	388	2	Melucci	Melucci	NNP
work_55utqx7tjrft5ojtbr67ypjdye	388	3	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	388	4	Nicola	Nicola	NNP
work_55utqx7tjrft5ojtbr67ypjdye	388	5	Orio	Orio	NNP
work_55utqx7tjrft5ojtbr67ypjdye	388	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	389	1	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	389	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	390	1	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	390	2	novel	novel	JJ
work_55utqx7tjrft5ojtbr67ypjdye	390	3	method	method	NN
work_55utqx7tjrft5ojtbr67ypjdye	390	4	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	390	5	stemmer	stemmer	JJ
work_55utqx7tjrft5ojtbr67ypjdye	390	6	generation	generation	NN
work_55utqx7tjrft5ojtbr67ypjdye	390	7	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	390	8	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	390	9	hidden	hidden	JJ
work_55utqx7tjrft5ojtbr67ypjdye	390	10	markov	markov	NN
work_55utqx7tjrft5ojtbr67ypjdye	390	11	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	390	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	391	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	391	2	Proceedings	proceeding	NNS
work_55utqx7tjrft5ojtbr67ypjdye	391	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	391	4	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	391	5	12th	12th	JJ
work_55utqx7tjrft5ojtbr67ypjdye	391	6	Inter-	Inter-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	391	7	national	national	JJ
work_55utqx7tjrft5ojtbr67ypjdye	391	8	Conference	Conference	NNP
work_55utqx7tjrft5ojtbr67ypjdye	391	9	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	391	10	Information	Information	NNP
work_55utqx7tjrft5ojtbr67ypjdye	391	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	391	12	Knowledge	Knowledge	NNP
work_55utqx7tjrft5ojtbr67ypjdye	391	13	Management	Management	NNP
work_55utqx7tjrft5ojtbr67ypjdye	391	14	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	391	15	CIKM	CIKM	NNP
work_55utqx7tjrft5ojtbr67ypjdye	391	16	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	391	17	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	391	18	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	391	19	131–138	131–138	CD
work_55utqx7tjrft5ojtbr67ypjdye	391	20	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	392	1	ACM	ACM	NNP
work_55utqx7tjrft5ojtbr67ypjdye	392	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	393	1	David	David	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	2	Mimno	Mimno	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	393	4	Hanna	Hanna	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	5	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	6	Wallach	Wallach	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	393	8	Edmund	Edmund	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	9	Talley	Talley	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	393	11	Miriam	Miriam	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	12	Leenders	Leenders	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	393	14	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	393	15	Andrew	Andrew	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	16	McCallum	McCallum	NNP
work_55utqx7tjrft5ojtbr67ypjdye	393	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	394	1	2011	2011	CD
work_55utqx7tjrft5ojtbr67ypjdye	394	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	395	1	Op-	op-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	395	2	timizing	timize	VBG
work_55utqx7tjrft5ojtbr67ypjdye	395	3	semantic	semantic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	395	4	coherence	coherence	NN
work_55utqx7tjrft5ojtbr67ypjdye	395	5	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	395	6	topic	topic	NN
work_55utqx7tjrft5ojtbr67ypjdye	395	7	models	model	NNS
work_55utqx7tjrft5ojtbr67ypjdye	395	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	396	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	396	2	Pro-	Pro-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	396	3	ceedings	ceeding	NNS
work_55utqx7tjrft5ojtbr67ypjdye	396	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	396	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	396	6	Conference	Conference	NNP
work_55utqx7tjrft5ojtbr67ypjdye	396	7	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	396	8	Empirical	Empirical	NNP
work_55utqx7tjrft5ojtbr67ypjdye	396	9	Methods	Methods	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	396	10	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	396	11	Natural	Natural	NNP
work_55utqx7tjrft5ojtbr67ypjdye	396	12	Language	Language	NNP
work_55utqx7tjrft5ojtbr67ypjdye	396	13	Processing	Processing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	396	14	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	396	15	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	396	16	262–272	262–272	CD
work_55utqx7tjrft5ojtbr67ypjdye	396	17	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	397	1	Asso-	asso-	IN
work_55utqx7tjrft5ojtbr67ypjdye	397	2	ciation	ciation	NN
work_55utqx7tjrft5ojtbr67ypjdye	397	3	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	397	4	Computational	Computational	NNP
work_55utqx7tjrft5ojtbr67ypjdye	397	5	Linguistics	Linguistics	NNP
work_55utqx7tjrft5ojtbr67ypjdye	397	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	398	1	Yuhong	Yuhong	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	2	Nan	Nan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	398	4	Min	Min	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	5	Yang	Yang	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	398	7	Zhemin	Zhemin	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	8	Yang	Yang	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	398	10	Shunfan	Shunfan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	11	Zhou	Zhou	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	12	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	398	13	Guofei	Guofei	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	14	Gu	Gu	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	15	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	398	16	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	398	17	XiaoFeng	XiaoFeng	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	18	Wang	Wang	NNP
work_55utqx7tjrft5ojtbr67ypjdye	398	19	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	399	1	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	399	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	400	1	Uipicker	uipicker	NN
work_55utqx7tjrft5ojtbr67ypjdye	400	2	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	400	3	User	user	NN
work_55utqx7tjrft5ojtbr67ypjdye	400	4	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	400	5	input	input	NN
work_55utqx7tjrft5ojtbr67ypjdye	400	6	privacy	privacy	NN
work_55utqx7tjrft5ojtbr67ypjdye	400	7	identification	identification	NN
work_55utqx7tjrft5ojtbr67ypjdye	400	8	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	400	9	mobile	mobile	NNP
work_55utqx7tjrft5ojtbr67ypjdye	400	10	applica-	applica-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	400	11	tions	tion	NNS
work_55utqx7tjrft5ojtbr67ypjdye	400	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	401	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	401	2	24th	24th	JJ
work_55utqx7tjrft5ojtbr67ypjdye	401	3	USENIX	USENIX	NNP
work_55utqx7tjrft5ojtbr67ypjdye	401	4	Security	Security	NNP
work_55utqx7tjrft5ojtbr67ypjdye	401	5	Symposium	Symposium	NNP
work_55utqx7tjrft5ojtbr67ypjdye	401	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	401	7	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	401	8	993–1008	993–1008	CD
work_55utqx7tjrft5ojtbr67ypjdye	401	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	402	1	Chris	Chris	NNP
work_55utqx7tjrft5ojtbr67ypjdye	402	2	D	D	NNP
work_55utqx7tjrft5ojtbr67ypjdye	402	3	Paice	Paice	NNP
work_55utqx7tjrft5ojtbr67ypjdye	402	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	403	1	1990	1990	CD
work_55utqx7tjrft5ojtbr67ypjdye	403	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	404	1	Another	another	DT
work_55utqx7tjrft5ojtbr67ypjdye	404	2	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	404	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	405	1	ACM	ACM	NNP
work_55utqx7tjrft5ojtbr67ypjdye	405	2	SIGIR	SIGIR	NNP
work_55utqx7tjrft5ojtbr67ypjdye	405	3	Forum	Forum	NNP
work_55utqx7tjrft5ojtbr67ypjdye	405	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	405	5	24(3):56–61	24(3):56–61	CD
work_55utqx7tjrft5ojtbr67ypjdye	405	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	406	1	299	299	CD
work_55utqx7tjrft5ojtbr67ypjdye	406	2	Martin	Martin	NNP
work_55utqx7tjrft5ojtbr67ypjdye	406	3	F	F	NNP
work_55utqx7tjrft5ojtbr67ypjdye	406	4	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	406	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	407	1	1980	1980	CD
work_55utqx7tjrft5ojtbr67ypjdye	407	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	408	1	An	an	DT
work_55utqx7tjrft5ojtbr67ypjdye	408	2	algorithm	algorithm	NN
work_55utqx7tjrft5ojtbr67ypjdye	408	3	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	408	4	suffix	suffix	NN
work_55utqx7tjrft5ojtbr67ypjdye	408	5	stripping	stripping	NN
work_55utqx7tjrft5ojtbr67ypjdye	408	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	409	1	Program	program	NN
work_55utqx7tjrft5ojtbr67ypjdye	409	2	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	409	3	14(3):130–137	14(3):130–137	CD
work_55utqx7tjrft5ojtbr67ypjdye	409	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	410	1	Martin	Martin	NNP
work_55utqx7tjrft5ojtbr67ypjdye	410	2	F	F	NNP
work_55utqx7tjrft5ojtbr67ypjdye	410	3	Porter	Porter	NNP
work_55utqx7tjrft5ojtbr67ypjdye	410	4	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	411	1	2001	2001	CD
work_55utqx7tjrft5ojtbr67ypjdye	411	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	412	1	Snowball	snowball	NN
work_55utqx7tjrft5ojtbr67ypjdye	412	2	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	412	3	A	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	412	4	lan-	lan-	NN
work_55utqx7tjrft5ojtbr67ypjdye	412	5	guage	guage	NN
work_55utqx7tjrft5ojtbr67ypjdye	412	6	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	412	7	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	412	8	algorithms	algorithm	NNS
work_55utqx7tjrft5ojtbr67ypjdye	412	9	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	413	1	Available	available	JJ
work_55utqx7tjrft5ojtbr67ypjdye	413	2	at	at	IN
work_55utqx7tjrft5ojtbr67ypjdye	413	3	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	413	4	http://www.snowball.tartarus.org/texts/introduction.html	http://www.snowball.tartarus.org/texts/introduction.html	NNP
work_55utqx7tjrft5ojtbr67ypjdye	413	5	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	414	1	SP	SP	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	2	Ruba	Ruba	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	3	Rani	Rani	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	414	5	B	B	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	6	Ramesh	Ramesh	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	7	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	414	8	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	9	Anusha	Anusha	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	414	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	414	12	JGR	JGR	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	13	Sathi-	Sathi-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	414	14	aseelan	aseelan	NN
work_55utqx7tjrft5ojtbr67ypjdye	414	15	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	415	1	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	415	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	416	1	Evaluation	evaluation	NN
work_55utqx7tjrft5ojtbr67ypjdye	416	2	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	416	3	stemming	stem	VBG
work_55utqx7tjrft5ojtbr67ypjdye	416	4	techniques	technique	NNS
work_55utqx7tjrft5ojtbr67ypjdye	416	5	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	416	6	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	416	7	classification	classification	NN
work_55utqx7tjrft5ojtbr67ypjdye	416	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	417	1	International	International	NNP
work_55utqx7tjrft5ojtbr67ypjdye	417	2	Journal	Journal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	417	3	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	417	4	Computer	Computer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	417	5	Science	Science	NNP
work_55utqx7tjrft5ojtbr67ypjdye	417	6	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	417	7	Mobile	Mobile	NNP
work_55utqx7tjrft5ojtbr67ypjdye	417	8	Computing	Computing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	417	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	417	10	4(3):165–171	4(3):165–171	CD
work_55utqx7tjrft5ojtbr67ypjdye	417	11	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	418	1	Evan	Evan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	418	2	Sandhaus	Sandhaus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	418	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	419	1	2008	2008	CD
work_55utqx7tjrft5ojtbr67ypjdye	419	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	420	1	The	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	420	2	new	new	NNP
work_55utqx7tjrft5ojtbr67ypjdye	420	3	york	york	NNP
work_55utqx7tjrft5ojtbr67ypjdye	420	4	times	times	NNP
work_55utqx7tjrft5ojtbr67ypjdye	420	5	anno-	anno-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	420	6	tated	tat	VBN
work_55utqx7tjrft5ojtbr67ypjdye	420	7	corpus	corpus	NNP
work_55utqx7tjrft5ojtbr67ypjdye	420	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	421	1	Linguistic	Linguistic	NNP
work_55utqx7tjrft5ojtbr67ypjdye	421	2	Data	Data	NNP
work_55utqx7tjrft5ojtbr67ypjdye	421	3	Consortium	Consortium	NNP
work_55utqx7tjrft5ojtbr67ypjdye	421	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	421	5	DVD	dvd	NN
work_55utqx7tjrft5ojtbr67ypjdye	421	6	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	421	7	LDC2009T19	ldc2009t19	NN
work_55utqx7tjrft5ojtbr67ypjdye	421	8	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	422	1	Ivan	Ivan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	422	2	Stankov	Stankov	NNP
work_55utqx7tjrft5ojtbr67ypjdye	422	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	422	4	Diman	Diman	NNP
work_55utqx7tjrft5ojtbr67ypjdye	422	5	Todorov	Todorov	NNP
work_55utqx7tjrft5ojtbr67ypjdye	422	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	422	7	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	422	8	Rossitza	Rossitza	NNP
work_55utqx7tjrft5ojtbr67ypjdye	422	9	Setchi	Setchi	NNP
work_55utqx7tjrft5ojtbr67ypjdye	422	10	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	423	1	2013	2013	CD
work_55utqx7tjrft5ojtbr67ypjdye	423	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	424	1	Enhanced	enhanced	JJ
work_55utqx7tjrft5ojtbr67ypjdye	424	2	cross	cross	JJ
work_55utqx7tjrft5ojtbr67ypjdye	424	3	-	-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	424	4	domain	domain	JJ
work_55utqx7tjrft5ojtbr67ypjdye	424	5	document	document	NN
work_55utqx7tjrft5ojtbr67ypjdye	424	6	clustering	cluster	VBG
work_55utqx7tjrft5ojtbr67ypjdye	424	7	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	424	8	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	424	9	semantically	semantically	RB
work_55utqx7tjrft5ojtbr67ypjdye	424	10	enhanced	enhance	VBN
work_55utqx7tjrft5ojtbr67ypjdye	424	11	text	text	NN
work_55utqx7tjrft5ojtbr67ypjdye	424	12	stemmer	stemmer	NN
work_55utqx7tjrft5ojtbr67ypjdye	424	13	(	(	-LRB-
work_55utqx7tjrft5ojtbr67ypjdye	424	14	sets	set	NNS
work_55utqx7tjrft5ojtbr67ypjdye	424	15	)	)	-RRB-
work_55utqx7tjrft5ojtbr67ypjdye	424	16	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	425	1	In-	In-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	425	2	ternational	ternational	JJ
work_55utqx7tjrft5ojtbr67ypjdye	425	3	Journal	Journal	NNP
work_55utqx7tjrft5ojtbr67ypjdye	425	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	425	5	Knowledge	Knowledge	NNP
work_55utqx7tjrft5ojtbr67ypjdye	425	6	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	425	7	based	base	VBN
work_55utqx7tjrft5ojtbr67ypjdye	425	8	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	425	9	Intelli-	Intelli-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	425	10	gent	gent	NN
work_55utqx7tjrft5ojtbr67ypjdye	425	11	Engineering	Engineering	NNP
work_55utqx7tjrft5ojtbr67ypjdye	425	12	Systems	Systems	NNPS
work_55utqx7tjrft5ojtbr67ypjdye	425	13	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	425	14	17(2):113–126	17(2):113–126	CD
work_55utqx7tjrft5ojtbr67ypjdye	425	15	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	426	1	Chuan	Chuan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	426	2	Su	Su	NNP
work_55utqx7tjrft5ojtbr67ypjdye	426	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	427	1	2015	2015	CD
work_55utqx7tjrft5ojtbr67ypjdye	427	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	428	1	Machine	machine	NN
work_55utqx7tjrft5ojtbr67ypjdye	428	2	learning	learn	VBG
work_55utqx7tjrft5ojtbr67ypjdye	428	3	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	428	4	reducing	reduce	VBG
work_55utqx7tjrft5ojtbr67ypjdye	428	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	428	6	ef-	ef-	JJ
work_55utqx7tjrft5ojtbr67ypjdye	428	7	fort	fort	NN
work_55utqx7tjrft5ojtbr67ypjdye	428	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	428	9	conducting	conduct	VBG
work_55utqx7tjrft5ojtbr67ypjdye	428	10	systematic	systematic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	428	11	reviews	review	NNS
work_55utqx7tjrft5ojtbr67ypjdye	428	12	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	428	13	SE	SE	NNP
work_55utqx7tjrft5ojtbr67ypjdye	428	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	429	1	Bachelor	Bachelor	NNP
work_55utqx7tjrft5ojtbr67ypjdye	429	2	Thesis	Thesis	NNP
work_55utqx7tjrft5ojtbr67ypjdye	429	3	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	430	1	Kristina	Kristina	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	2	Toutanova	Toutanova	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	3	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	430	4	Dan	Dan	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	5	Klein	Klein	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	6	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	430	7	Christopher	Christopher	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	8	D	D	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	9	Manning	Manning	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	10	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	430	11	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	430	12	Yoram	Yoram	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	13	Singer	Singer	NNP
work_55utqx7tjrft5ojtbr67ypjdye	430	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	431	1	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	431	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	432	1	Feature	feature	NN
work_55utqx7tjrft5ojtbr67ypjdye	432	2	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	432	3	rich	rich	JJ
work_55utqx7tjrft5ojtbr67ypjdye	432	4	part	part	NN
work_55utqx7tjrft5ojtbr67ypjdye	432	5	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	432	6	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	432	7	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	432	8	speech	speech	NN
work_55utqx7tjrft5ojtbr67ypjdye	432	9	tagging	tagging	NN
work_55utqx7tjrft5ojtbr67ypjdye	432	10	with	with	IN
work_55utqx7tjrft5ojtbr67ypjdye	432	11	a	a	DT
work_55utqx7tjrft5ojtbr67ypjdye	432	12	cyclic	cyclic	JJ
work_55utqx7tjrft5ojtbr67ypjdye	432	13	dependency	dependency	NN
work_55utqx7tjrft5ojtbr67ypjdye	432	14	network	network	NN
work_55utqx7tjrft5ojtbr67ypjdye	432	15	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	433	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	433	2	Pro-	Pro-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	3	ceedings	ceeding	NNS
work_55utqx7tjrft5ojtbr67ypjdye	433	4	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	433	5	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	433	6	2003	2003	CD
work_55utqx7tjrft5ojtbr67ypjdye	433	7	Conference	Conference	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	8	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	433	9	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	433	10	North	North	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	11	Ameri-	Ameri-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	12	can	can	MD
work_55utqx7tjrft5ojtbr67ypjdye	433	13	Chapter	chapter	NN
work_55utqx7tjrft5ojtbr67ypjdye	433	14	of	of	IN
work_55utqx7tjrft5ojtbr67ypjdye	433	15	the	the	DT
work_55utqx7tjrft5ojtbr67ypjdye	433	16	Association	Association	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	17	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	433	18	Computational	Computational	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	19	Lin-	Lin-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	20	guistics	guistic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	433	21	on	on	IN
work_55utqx7tjrft5ojtbr67ypjdye	433	22	Human	Human	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	23	Language	Language	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	24	Technology	Technology	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	25	-	-	HYPH
work_55utqx7tjrft5ojtbr67ypjdye	433	26	Volume	Volume	NNP
work_55utqx7tjrft5ojtbr67ypjdye	433	27	1	1	CD
work_55utqx7tjrft5ojtbr67ypjdye	433	28	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	433	29	pages	page	VBZ
work_55utqx7tjrft5ojtbr67ypjdye	433	30	173–180	173–180	CD
work_55utqx7tjrft5ojtbr67ypjdye	433	31	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	434	1	Association	Association	NNP
work_55utqx7tjrft5ojtbr67ypjdye	434	2	for	for	IN
work_55utqx7tjrft5ojtbr67ypjdye	434	3	Computational	Computational	NNP
work_55utqx7tjrft5ojtbr67ypjdye	434	4	Lin-	Lin-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	434	5	guistics	guistic	NNS
work_55utqx7tjrft5ojtbr67ypjdye	434	6	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	435	1	Hanna	Hanna	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	2	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	3	Wallach	Wallach	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	4	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	435	5	David	David	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	6	M	M	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	7	Mimno	Mimno	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	8	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	435	9	and	and	CC
work_55utqx7tjrft5ojtbr67ypjdye	435	10	Andrew	Andrew	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	11	K	K	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	12	Mc-	Mc-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	13	Callum	Callum	NNP
work_55utqx7tjrft5ojtbr67ypjdye	435	14	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	436	1	2009	2009	CD
work_55utqx7tjrft5ojtbr67ypjdye	436	2	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	437	1	Rethinking	rethink	VBG
work_55utqx7tjrft5ojtbr67ypjdye	437	2	LDA	LDA	NNP
work_55utqx7tjrft5ojtbr67ypjdye	437	3	:	:	:
work_55utqx7tjrft5ojtbr67ypjdye	437	4	Why	why	WRB
work_55utqx7tjrft5ojtbr67ypjdye	437	5	priors	prior	NNS
work_55utqx7tjrft5ojtbr67ypjdye	437	6	matter	matter	VBP
work_55utqx7tjrft5ojtbr67ypjdye	437	7	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	438	1	In	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	438	2	Advances	advance	NNS
work_55utqx7tjrft5ojtbr67ypjdye	438	3	in	in	IN
work_55utqx7tjrft5ojtbr67ypjdye	438	4	Neural	Neural	NNP
work_55utqx7tjrft5ojtbr67ypjdye	438	5	Information	Information	NNP
work_55utqx7tjrft5ojtbr67ypjdye	438	6	Processing	Processing	NNP
work_55utqx7tjrft5ojtbr67ypjdye	438	7	Sys-	Sys-	NNP
work_55utqx7tjrft5ojtbr67ypjdye	438	8	tems	tem	NNS
work_55utqx7tjrft5ojtbr67ypjdye	438	9	,	,	,
work_55utqx7tjrft5ojtbr67ypjdye	438	10	pages	page	NNS
work_55utqx7tjrft5ojtbr67ypjdye	438	11	1973–1981	1973–1981	CD
work_55utqx7tjrft5ojtbr67ypjdye	438	12	.	.	.
work_55utqx7tjrft5ojtbr67ypjdye	439	1	300	300	CD