id	sid	tid	token	lemma	pos
5767	1	1	lib	lib	NNP
5767	1	2	-	-	HYPH
5767	1	3	s	s	NNP
5767	1	4	-	-	HYPH
5767	1	5	mocs	mocs	NNP
5767	1	6	-	-	HYPH
5767	1	7	kmc364	kmc364	NN
5767	1	8	-	-	HYPH
5767	1	9	20141005044052	20141005044052	CD
5767	1	10	109	109	CD
5767	1	11	Statistical	statistical	JJ
5767	1	12	Behavior	Behavior	NNP
5767	1	13	of	of	IN
5767	1	14	Search	Search	NNP
5767	1	15	Keys	Keys	NNPS
5767	1	16	Abraham	Abraham	NNP
5767	1	17	BOOKSTEIN	BOOKSTEIN	NNP
5767	1	18	:	:	:
5767	1	19	Graduate	Graduate	NNP
5767	1	20	Library	Library	NNP
5767	1	21	School	School	NNP
5767	1	22	,	,	,
5767	1	23	University	University	NNP
5767	1	24	of	of	IN
5767	1	25	Chicago	Chicago	NNP
5767	1	26	Editor	Editor	NNP
5767	1	27	's	's	POS
5767	1	28	note	note	NN
5767	1	29	:	:	:
5767	1	30	The	the	DT
5767	1	31	editor	editor	NN
5767	1	32	and	and	CC
5767	1	33	author	author	NN
5767	1	34	are	be	VBP
5767	1	35	aware	aware	JJ
5767	1	36	that	that	IN
5767	1	37	varying	vary	VBG
5767	1	38	approaches	approach	NNS
5767	1	39	may	may	MD
5767	1	40	be	be	VB
5767	1	41	taken	take	VBN
5767	1	42	to	to	IN
5767	1	43	the	the	DT
5767	1	44	problem	problem	NN
5767	1	45	presented	present	VBN
5767	1	46	here	here	RB
5767	1	47	.	.	.
5767	2	1	Readers	reader	NNS
5767	2	2	are	be	VBP
5767	2	3	invited	invite	VBN
5767	2	4	to	to	TO
5767	2	5	respond	respond	VB
5767	2	6	in	in	IN
5767	2	7	the	the	DT
5767	2	8	form	form	NN
5767	2	9	of	of	IN
5767	2	10	a	a	DT
5767	2	11	paper	paper	NN
5767	2	12	or	or	CC
5767	2	13	a	a	DT
5767	2	14	technical	technical	JJ
5767	2	15	C.'Ommunication	C.'Ommunication	NNP
5767	2	16	.	.	.
5767	3	1	In	in	IN
5767	3	2	discussion	discussion	NN
5767	3	3	about	about	IN
5767	3	4	search	search	NN
5767	3	5	keys	key	NNS
5767	3	6	,	,	,
5767	3	7	concern	concern	NN
5767	3	8	has	have	VBZ
5767	3	9	been	be	VBN
5767	3	10	expressed	express	VBN
5767	3	11	as	as	IN
5767	3	12	to	to	IN
5767	3	13	how	how	WRB
5767	3	14	the	the	DT
5767	3	15	nwnber	nwnber	NN
5767	3	16	of	of	IN
5767	3	17	items	item	NNS
5767	3	18	tetrieved	tetrieve	VBN
5767	3	19	by	by	IN
5767	3	20	a	a	DT
5767	3	21	single	single	JJ
5767	3	22	value	value	NN
5767	3	23	relates	relate	VBZ
5767	3	24	to	to	IN
5767	3	25	collection	collection	NN
5767	3	26	size	size	NN
5767	3	27	.	.	.
5767	4	1	This	this	DT
5767	4	2	paper	paper	NN
5767	4	3	creates	create	VBZ
5767	4	4	a	a	DT
5767	4	5	statistical	statistical	JJ
5767	4	6	model	model	NN
5767	4	7	that	that	WDT
5767	4	8	attempts	attempt	VBZ
5767	4	9	to	to	TO
5767	4	10	give	give	VB
5767	4	11	some	some	DT
5767	4	12	insight	insight	NN
5767	4	13	into	into	IN
5767	4	14	this	this	DT
5767	4	15	behavior	behavior	NN
5767	4	16	.	.	.
5767	5	1	It	-PRON-	PRP
5767	5	2	is	be	VBZ
5767	5	3	concluded	conclude	VBN
5767	5	4	that	that	IN
5767	5	5	,	,	,
5767	5	6	in	in	IN
5767	5	7	general	general	JJ
5767	5	8	,	,	,
5767	5	9	the	the	DT
5767	5	10	observed	observed	JJ
5767	5	11	behavior	behavior	NN
5767	5	12	can	can	MD
5767	5	13	be	be	VB
5767	5	14	explained	explain	VBN
5767	5	15	as	as	IN
5767	5	16	being	be	VBG
5767	5	17	intrinsically	intrinsically	RB
5767	5	18	statistical	statistical	JJ
5767	5	19	in	in	IN
5767	5	20	nature	nature	NN
5767	5	21	rather	rather	RB
5767	5	22	than	than	IN
5767	5	23	being	be	VBG
5767	5	24	a	a	DT
5767	5	25	property	property	NN
5767	5	26	of	of	IN
5767	5	27	specific	specific	JJ
5767	5	28	search	search	NN
5767	5	29	keys	key	NNS
5767	5	30	.	.	.
5767	6	1	An	an	DT
5767	6	2	attempt	attempt	NN
5767	6	3	is	be	VBZ
5767	6	4	made	make	VBN
5767	6	5	to	to	TO
5767	6	6	relate	relate	VB
5767	6	7	this	this	DT
5767	6	8	model	model	NN
5767	6	9	to	to	IN
5767	6	10	other	other	JJ
5767	6	11	tesearch	tesearch	NN
5767	6	12	,	,	,
5767	6	13	and	and	CC
5767	6	14	to	to	TO
5767	6	15	indicate	indicate	VB
5767	6	16	how	how	WRB
5767	6	17	this	this	DT
5767	6	18	model	model	NN
5767	6	19	may	may	MD
5767	6	20	be	be	VB
5767	6	21	made	make	VBN
5767	6	22	to	to	TO
5767	6	23	yield	yield	VB
5767	6	24	more	more	RBR
5767	6	25	accurate	accurate	JJ
5767	6	26	predictions	prediction	NNS
5767	6	27	.	.	.
5767	7	1	INTRODUCTION	introduction	VB
5767	7	2	Various	various	JJ
5767	7	3	experiments	experiment	NNS
5767	7	4	suggest	suggest	VBP
5767	7	5	that	that	IN
5767	7	6	it	-PRON-	PRP
5767	7	7	may	may	MD
5767	7	8	be	be	VB
5767	7	9	possible	possible	JJ
5767	7	10	to	to	TO
5767	7	11	develop	develop	VB
5767	7	12	,	,	,
5767	7	13	as	as	IN
5767	7	14	an	an	DT
5767	7	15	access	access	NN
5767	7	16	route	route	NN
5767	7	17	into	into	IN
5767	7	18	a	a	DT
5767	7	19	file	file	NN
5767	7	20	of	of	IN
5767	7	21	bibliographic	bibliographic	JJ
5767	7	22	records	record	NNS
5767	7	23	,	,	,
5767	7	24	a	a	DT
5767	7	25	search	search	NN
5767	7	26	key	key	NN
5767	7	27	'	'	''
5767	7	28	"	"	''
5767	7	29	whose	whose	WP$
5767	7	30	values	value	NNS
5767	7	31	can	can	MD
5767	7	32	be	be	VB
5767	7	33	easily	easily	RB
5767	7	34	derived	derive	VBN
5767	7	35	from	from	IN
5767	7	36	such	such	JJ
5767	7	37	bibliographic	bibliographic	JJ
5767	7	38	data	datum	NNS
5767	7	39	as	as	IN
5767	7	40	is	be	VBZ
5767	7	41	likely	likely	JJ
5767	7	42	to	to	TO
5767	7	43	be	be	VB
5767	7	44	available	available	JJ
5767	7	45	to	to	IN
5767	7	46	its	-PRON-	PRP$
5767	7	47	users.1	users.1	CD
5767	7	48	Some	some	DT
5767	7	49	concern	concern	NN
5767	7	50	,	,	,
5767	7	51	however	however	RB
5767	7	52	,	,	,
5767	7	53	has	have	VBZ
5767	7	54	been	be	VBN
5767	7	55	expressed	express	VBN
5767	7	56	regarding	regard	VBG
5767	7	57	the	the	DT
5767	7	58	non-	non-	NN
5767	7	59	uniqueness	uniqueness	NN
5767	7	60	of	of	IN
5767	7	61	these	these	DT
5767	7	62	keys	key	NNS
5767	7	63	:	:	:
5767	7	64	if	if	IN
5767	7	65	the	the	DT
5767	7	66	number	number	NN
5767	7	67	of	of	IN
5767	7	68	items	item	NNS
5767	7	69	retrieved	retrieve	VBN
5767	7	70	were	be	VBD
5767	7	71	often	often	RB
5767	7	72	to	to	TO
5767	7	73	exceed	exceed	VB
5767	7	74	an	an	DT
5767	7	75	amount	amount	NN
5767	7	76	easily	easily	RB
5767	7	77	handled	handle	VBN
5767	7	78	by	by	IN
5767	7	79	a	a	DT
5767	7	80	user	user	NN
5767	7	81	of	of	IN
5767	7	82	the	the	DT
5767	7	83	system	system	NN
5767	7	84	,	,	,
5767	7	85	the	the	DT
5767	7	86	value	value	NN
5767	7	87	of	of	IN
5767	7	88	this	this	DT
5767	7	89	access	access	NN
5767	7	90	route	route	NN
5767	7	91	would	would	MD
5767	7	92	be	be	VB
5767	7	93	considerably	considerably	RB
5767	7	94	diminished	diminish	VBN
5767	7	95	.	.	.
5767	8	1	Accordingly	accordingly	RB
5767	8	2	,	,	,
5767	8	3	an	an	DT
5767	8	4	important	important	JJ
5767	8	5	measure	measure	NN
5767	8	6	of	of	IN
5767	8	7	search	search	NN
5767	8	8	key	key	JJ
5767	8	9	performance	performance	NN
5767	8	10	is	be	VBZ
5767	8	11	the	the	DT
5767	8	12	frequency	frequency	NN
5767	8	13	with	with	IN
5767	8	14	which	which	WDT
5767	8	15	a	a	DT
5767	8	16	large	large	JJ
5767	8	17	number	number	NN
5767	8	18	of	of	IN
5767	8	19	records	record	NNS
5767	8	20	is	be	VBZ
5767	8	21	reh·ieved	reh·ieve	VBN
5767	8	22	as	as	IN
5767	8	23	the	the	DT
5767	8	24	search	search	NN
5767	8	25	key	key	NN
5767	8	26	is	be	VBZ
5767	8	27	applied	apply	VBN
5767	8	28	to	to	IN
5767	8	29	the	the	DT
5767	8	30	file	file	NN
5767	8	31	.	.	.
5767	9	1	This	this	DT
5767	9	2	measure	measure	NN
5767	9	3	is	be	VBZ
5767	9	4	·	·	NFP
5767	9	5	related	relate	VBN
5767	9	6	,	,	,
5767	9	7	for	for	IN
5767	9	8	example	example	NN
5767	9	9	,	,	,
5767	9	10	to	to	IN
5767	9	11	how	how	WRB
5767	9	12	many	many	JJ
5767	9	13	memory	memory	NN
5767	9	14	accesses	access	NNS
5767	9	15	will	will	MD
5767	9	16	be	be	VB
5767	9	17	re-	re-	RB
5767	9	18	quired	quire	VBN
5767	9	19	,	,	,
5767	9	20	on	on	IN
5767	9	21	the	the	DT
5767	9	22	average	average	NN
5767	9	23	,	,	,
5767	9	24	to	to	TO
5767	9	25	retrieve	retrieve	VB
5767	9	26	all	all	DT
5767	9	27	records	record	NNS
5767	9	28	satisfying	satisfy	VBG
5767	9	29	a	a	DT
5767	9	30	request	request	NN
5767	9	31	;	;	:
5767	9	32	it	-PRON-	PRP
5767	9	33	is	be	VBZ
5767	9	34	also	also	RB
5767	9	35	an	an	DT
5767	9	36	important	important	JJ
5767	9	37	consideration	consideration	NN
5767	9	38	in	in	IN
5767	9	39	deciding	decide	VBG
5767	9	40	which	which	WDT
5767	9	41	display	display	NN
5767	9	42	device	device	NN
5767	9	43	should	should	MD
5767	9	44	be	be	VB
5767	9	45	in-	in-	RB
5767	9	46	stalled	stall	VBN
5767	9	47	in	in	IN
5767	9	48	a	a	DT
5767	9	49	system.2	system.2	NN
5767	9	50	•	•	NN
5767	9	51	3	3	CD
5767	9	52	After	after	IN
5767	9	53	evaluating	evaluate	VBG
5767	9	54	such	such	PDT
5767	9	55	a	a	DT
5767	9	56	measure	measure	NN
5767	9	57	for	for	IN
5767	9	58	a	a	DT
5767	9	59	search	search	NN
5767	9	60	key	key	NN
5767	9	61	on	on	IN
5767	9	62	a	a	DT
5767	9	63	particular	particular	JJ
5767	9	64	file	file	NN
5767	9	65	,	,	,
5767	9	66	it	-PRON-	PRP
5767	9	67	is	be	VBZ
5767	9	68	reasonable	reasonable	JJ
5767	9	69	to	to	TO
5767	9	70	ask	ask	VB
5767	9	71	how	how	WRB
5767	9	72	that	that	DT
5767	9	73	measure	measure	NN
5767	9	74	will	will	MD
5767	9	75	change	change	VB
5767	9	76	over	over	IN
5767	9	77	time	time	NN
5767	9	78	,	,	,
5767	9	79	as	as	IN
5767	9	80	the	the	DT
5767	9	81	file	file	NN
5767	9	82	in-	in-	NNP
5767	9	83	creases	crease	VBZ
5767	9	84	in	in	IN
5767	9	85	size	size	NN
5767	9	86	.	.	.
5767	10	1	The	the	DT
5767	10	2	nature	nature	NN
5767	10	3	of	of	IN
5767	10	4	this	this	DT
5767	10	5	variation	variation	NN
5767	10	6	has	have	VBZ
5767	10	7	already	already	RB
5767	10	8	been	be	VBN
5767	10	9	of	of	IN
5767	10	10	concern	concern	NN
5767	10	11	to	to	IN
5767	10	12	researchers	researcher	NNS
5767	10	13	in	in	IN
5767	10	14	the	the	DT
5767	10	15	field	field	NN
5767	10	16	.	.	.
5767	11	1	Kilgour	Kilgour	NNP
5767	11	2	,	,	,
5767	11	3	on	on	IN
5767	11	4	the	the	DT
5767	11	5	basis	basis	NN
5767	11	6	of	of	IN
5767	11	7	a	a	DT
5767	11	8	·	·	NFP
5767	11	9	number	number	NN
5767	11	10	of	of	IN
5767	11	11	experiments	experiment	NNS
5767	11	12	carried	carry	VBN
5767	11	13	out	out	RP
5767	11	14	at	at	IN
5767	11	15	OCLC	oclc	NN
5767	11	16	,	,	,
5767	11	17	notes	note	VBZ
5767	11	18	that	that	IN
5767	11	19	"	"	``
5767	11	20	There	there	EX
5767	11	21	remains	remain	VBZ
5767	11	22	a	a	DT
5767	11	23	major	major	JJ
5767	11	24	problem	problem	NN
5767	11	25	to	to	TO
5767	11	26	be	be	VB
5767	11	27	o	o	XX
5767	11	28	By	by	IN
5767	11	29	the	the	DT
5767	11	30	.	.	.
5767	12	1	phrase	phrase	VB
5767	12	2	"	"	``
5767	12	3	search	search	NN
5767	12	4	key'~	key'~	NNP
5767	12	5	we	-PRON-	PRP
5767	12	6	mean	mean	VBP
5767	12	7	a	a	DT
5767	12	8	key	key	NN
5767	12	9	similar	similar	JJ
5767	12	10	to	to	IN
5767	12	11	the	the	DT
5767	12	12	3	3	CD
5767	12	13	-	-	SYM
5767	12	14	3	3	CD
5767	12	15	or	or	CC
5767	12	16	3	3	CD
5767	12	17	-	-	HYPH
5767	12	18	1	1	CD
5767	12	19	-	-	HYPH
5767	12	20	1	1	CD
5767	12	21	-	-	HYPH
5767	12	22	1	1	CD
5767	12	23	keys	key	NNS
5767	12	24	used	use	VBN
5767	12	25	at	at	IN
5767	12	26	·	·	NFP
5767	12	27	Ohio	Ohio	NNP
5767	12	28	College	College	NNP
5767	12	29	Library	Library	NNP
5767	12	30	Center	Center	NNP
5767	12	31	and	and	CC
5767	12	32	other	other	JJ
5767	12	33	places	place	NNS
5767	12	34	,	,	,
5767	12	35	which	which	WDT
5767	12	36	is	be	VBZ
5767	12	37	made	make	VBN
5767	12	38	up	up	RP
5767	12	39	by	by	IN
5767	12	40	concatenating	concatenate	VBG
5767	12	41	truncations	truncation	NNS
5767	12	42	of	of	IN
5767	12	43	bibliographic	bibliographic	JJ
5767	12	44	data	datum	NNS
5767	12	45	elements	element	NNS
5767	12	46	.	.	.
5767	13	1	llO	llO	NNP
5767	13	2	Journal	Journal	NNP
5767	13	3	of	of	IN
5767	13	4	Library	Library	NNP
5767	13	5	Automation	Automation	NNP
5767	13	6	Vol	Vol	NNP
5767	13	7	6/	6/	CD
5767	13	8	2	2	CD
5767	13	9	June	June	NNP
5767	13	10	1973	1973	CD
5767	13	11	solved	solve	VBD
5767	13	12	and	and	CC
5767	13	13	a	a	DT
5767	13	14	major	major	JJ
5767	13	15	question	question	NN
5767	13	16	to	to	TO
5767	13	17	be	be	VB
5767	13	18	answered	answer	VBN
5767	13	19	.	.	.
5767	14	1	The	the	DT
5767	14	2	problem	problem	NN
5767	14	3	is	be	VBZ
5767	14	4	constituted	constituted	JJ
5767	14	5	of	of	IN
5767	14	6	those	those	DT
5767	14	7	replies	reply	NNS
5767	14	8	that	that	WDT
5767	14	9	contain	contain	VBP
5767	14	10	a	a	DT
5767	14	11	number	number	NN
5767	14	12	of	of	IN
5767	14	13	entries	entry	NNS
5767	14	14	exceeding	exceed	VBG
5767	14	15	the	the	DT
5767	14	16	optimal	optimal	NNP
5767	14	17	maximum	maximum	NN
5767	14	18	..	..	.
5767	14	19	..	..	.
5767	15	1	The	the	DT
5767	15	2	major	major	JJ
5767	15	3	question	question	NN
5767	15	4	to	to	TO
5767	15	5	be	be	VB
5767	15	6	answered	answer	VBN
5767	15	7	is	be	VBZ
5767	15	8	how	how	WRB
5767	15	9	truncated	truncate	VBN
5767	15	10	search	search	NN
5767	15	11	keys	key	NNS
5767	15	12	will	will	MD
5767	15	13	perform	perform	VB
5767	15	14	on	on	IN
5767	15	15	files	file	NNS
5767	15	16	ten	ten	CD
5767	15	17	and	and	CC
5767	15	18	a	a	DT
5767	15	19	hundred	hundred	CD
5767	15	20	times	time	NNS
5767	15	21	the	the	DT
5767	15	22	size	size	NN
5767	15	23	of	of	IN
5767	15	24	that	that	DT
5767	15	25	used	use	VBN
5767	15	26	in	in	IN
5767	15	27	this	this	DT
5767	15	28	experiment	experiment	NN
5767	15	29	.	.	.
5767	15	30	"	"	''
5767	15	31	'	'	''
5767	16	1	He	-PRON-	PRP
5767	16	2	elsewhere	elsewhere	RB
5767	16	3	observes	observe	VBZ
5767	16	4	that	that	IN
5767	16	5	"	"	``
5767	16	6	as	as	IN
5767	16	7	a	a	DT
5767	16	8	file	file	NN
5767	16	9	of	of	IN
5767	16	10	bibliographic	bibliographic	JJ
5767	16	11	entries	entry	NNS
5767	16	12	increases	increase	NNS
5767	16	13	,	,	,
5767	16	14	the	the	DT
5767	16	15	maximum	maximum	JJ
5767	16	16	number	number	NN
5767	16	17	of	of	IN
5767	16	18	entries	entry	NNS
5767	16	19	per	per	IN
5767	16	20	reply	reply	NN
5767	16	21	does	do	VBZ
5767	16	22	not	not	RB
5767	16	23	in-	in-	JJ
5767	16	24	crease	crease	NN
5767	16	25	in	in	IN
5767	16	26	a	a	DT
5767	16	27	one	one	CD
5767	16	28	-	-	HYPH
5767	16	29	to	to	IN
5767	16	30	-	-	HYPH
5767	16	31	one	one	CD
5767	16	32	ratio	ratio	NN
5767	16	33	...	...	:
5767	16	34	.	.	.
5767	17	1	"	"	``
5767	17	2	5	5	CD
5767	17	3	This	this	DT
5767	17	4	paper	paper	NN
5767	17	5	presents	present	VBZ
5767	17	6	a	a	DT
5767	17	7	mathematical	mathematical	JJ
5767	17	8	model	model	NN
5767	17	9	that	that	WDT
5767	17	10	addresses	address	VBZ
5767	17	11	itself	-PRON-	PRP
5767	17	12	to	to	IN
5767	17	13	the	the	DT
5767	17	14	problem	problem	NN
5767	17	15	defined	define	VBN
5767	17	16	by	by	IN
5767	17	17	Kilgour	Kilgour	NNP
5767	17	18	and	and	CC
5767	17	19	attempts	attempt	VBZ
5767	17	20	to	to	TO
5767	17	21	explain	explain	VB
5767	17	22	his	-PRON-	PRP$
5767	17	23	observation	observation	NN
5767	17	24	;	;	:
5767	17	25	it	-PRON-	PRP
5767	17	26	is	be	VBZ
5767	17	27	suggested	suggest	VBN
5767	17	28	that	that	IN
5767	17	29	the	the	DT
5767	17	30	gross	gross	JJ
5767	17	31	features	feature	NNS
5767	17	32	of	of	IN
5767	17	33	the	the	DT
5767	17	34	be-	be-	NNP
5767	17	35	havior	havior	NNP
5767	17	36	are	be	VBP
5767	17	37	statistical	statistical	JJ
5767	17	38	in	in	IN
5767	17	39	nature	nature	NN
5767	17	40	and	and	CC
5767	17	41	not	not	RB
5767	17	42	properties	property	NNS
5767	17	43	of	of	IN
5767	17	44	specific	specific	JJ
5767	17	45	search	search	NN
5767	17	46	keys	key	NNS
5767	17	47	.	.	.
5767	18	1	A	a	DT
5767	18	2	VIEW	view	NN
5767	18	3	OF	of	IN
5767	18	4	COLLECTION	collection	NN
5767	18	5	GROWTH	GROWTH	NNS
5767	18	6	The	the	DT
5767	18	7	cause	cause	NN
5767	18	8	of	of	IN
5767	18	9	the	the	DT
5767	18	10	phenomenon	phenomenon	NN
5767	18	11	observed	observe	VBN
5767	18	12	by	by	IN
5767	18	13	Kilgour	Kilgour	NNP
5767	18	14	can	can	MD
5767	18	15	best	best	RB
5767	18	16	be	be	VB
5767	18	17	under-	under-	RB
5767	18	18	stood	stand	VBD
5767	18	19	by	by	IN
5767	18	20	first	first	RB
5767	18	21	considering	consider	VBG
5767	18	22	a	a	DT
5767	18	23	simple	simple	JJ
5767	18	24	model	model	NN
5767	18	25	which	which	WDT
5767	18	26	,	,	,
5767	18	27	while	while	IN
5767	18	28	not	not	RB
5767	18	29	itself	-PRON-	PRP
5767	18	30	valid	valid	JJ
5767	18	31	,	,	,
5767	18	32	does	do	VBZ
5767	18	33	cast	cast	VB
5767	18	34	light	light	NN
5767	18	35	on	on	IN
5767	18	36	the	the	DT
5767	18	37	nature	nature	NN
5767	18	38	of	of	IN
5767	18	39	the	the	DT
5767	18	40	behavior	behavior	NN
5767	18	41	.	.	.
5767	19	1	This	this	DT
5767	19	2	first	first	JJ
5767	19	3	model	model	NN
5767	19	4	neglects	neglect	VBZ
5767	19	5	the	the	DT
5767	19	6	effect	effect	NN
5767	19	7	of	of	IN
5767	19	8	randomness	randomness	NN
5767	19	9	both	both	DT
5767	19	10	in	in	IN
5767	19	11	the	the	DT
5767	19	12	growth	growth	NN
5767	19	13	of	of	IN
5767	19	14	the	the	DT
5767	19	15	collection	collection	NN
5767	19	16	and	and	CC
5767	19	17	in	in	IN
5767	19	18	the	the	DT
5767	19	19	arrival	arrival	NN
5767	19	20	of	of	IN
5767	19	21	requests	request	NNS
5767	19	22	.	.	.
5767	20	1	It	-PRON-	PRP
5767	20	2	supposes	suppose	VBZ
5767	20	3	our	-PRON-	PRP$
5767	20	4	search	search	NN
5767	20	5	key	key	NN
5767	20	6	has	have	VBZ
5767	20	7	the	the	DT
5767	20	8	following	follow	VBG
5767	20	9	property	property	NN
5767	20	10	:	:	:
5767	20	11	regardless	regardless	RB
5767	20	12	of	of	IN
5767	20	13	collection	collection	NN
5767	20	14	size	size	NN
5767	20	15	,	,	,
5767	20	16	the	the	DT
5767	20	17	fraction	fraction	NN
5767	20	18	of	of	IN
5767	20	19	the	the	DT
5767	20	20	collection	collection	NN
5767	20	21	retrieved	retrieve	VBN
5767	20	22	by	by	IN
5767	20	23	a	a	DT
5767	20	24	particular	particular	JJ
5767	20	25	search	search	NN
5767	20	26	key	key	JJ
5767	20	27	value	value	NN
5767	20	28	,	,	,
5767	20	29	v~	v~	NNP
5767	20	30	,	,	,
5767	20	31	is	be	VBZ
5767	20	32	exactly	exactly	RB
5767	20	33	given	give	VBN
5767	20	34	by	by	IN
5767	20	35	a	a	DT
5767	20	36	constant	constant	JJ
5767	20	37	f	f	NN
5767	20	38	;	;	:
5767	20	39	;	;	:
5767	20	40	thus	thus	RB
5767	20	41	,	,	,
5767	20	42	if	if	IN
5767	20	43	the	the	DT
5767	20	44	fil	fil	NNP
5767	20	45	e	e	NNP
5767	20	46	holds	hold	VBZ
5767	20	47	N	N	NNP
5767	20	48	records	record	NNS
5767	20	49	,	,	,
5767	20	50	a	a	DT
5767	20	51	request	request	NN
5767	20	52	for	for	IN
5767	20	53	v	v	NN
5767	20	54	1	1	CD
5767	20	55	will	will	MD
5767	20	56	retrieve	retrieve	VB
5767	20	57	n	n	CC
5767	20	58	1	1	CD
5767	20	59	=	=	SYM
5767	20	60	f	f	NN
5767	20	61	,	,	,
5767	20	62	N	N	NNP
5767	20	63	records	record	NNS
5767	20	64	.	.	.
5767	21	1	This	this	DT
5767	21	2	model	model	NN
5767	21	3	similarly	similarly	RB
5767	21	4	assumes	assume	VBZ
5767	21	5	that	that	IN
5767	21	6	among	among	IN
5767	21	7	any	any	DT
5767	21	8	sizeable	sizeable	JJ
5767	21	9	number	number	NN
5767	21	10	of	of	IN
5767	21	11	requests	request	NNS
5767	21	12	,	,	,
5767	21	13	the	the	DT
5767	21	14	fraction	fraction	NN
5767	21	15	of	of	IN
5767	21	16	the	the	DT
5767	21	17	time	time	NN
5767	21	18	any	any	DT
5767	21	19	particular	particular	JJ
5767	21	20	search	search	NN
5767	21	21	key	key	JJ
5767	21	22	value	value	NN
5767	21	23	will	will	MD
5767	21	24	occur	occur	VB
5767	21	25	is	be	VBZ
5767	21	26	fixed	fix	VBN
5767	21	27	;	;	:
5767	21	28	thus	thus	RB
5767	21	29	,	,	,
5767	21	30	for	for	IN
5767	21	31	any	any	DT
5767	21	32	subset	subset	NN
5767	21	33	of	of	IN
5767	21	34	search	search	NN
5767	21	35	key	key	JJ
5767	21	36	values	value	NNS
5767	21	37	,	,	,
5767	21	38	it	-PRON-	PRP
5767	21	39	is	be	VBZ
5767	21	40	possible	possible	JJ
5767	21	41	to	to	TO
5767	21	42	determine	determine	VB
5767	21	43	how	how	WRB
5767	21	44	often	often	RB
5767	21	45	members	member	NNS
5767	21	46	of	of	IN
5767	21	47	that	that	DT
5767	21	48	subset	subset	NN
5767	21	49	will	will	MD
5767	21	50	occur	occur	VB
5767	21	51	among	among	IN
5767	21	52	a	a	DT
5767	21	53	set	set	NN
5767	21	54	of	of	IN
5767	21	55	requests	request	NNS
5767	21	56	.	.	.
5767	22	1	In	in	IN
5767	22	2	particular	particular	JJ
5767	22	3	,	,	,
5767	22	4	for	for	IN
5767	22	5	any	any	DT
5767	22	6	integer	integer	NN
5767	22	7	n	n	NN
5767	22	8	,	,	,
5767	22	9	we	-PRON-	PRP
5767	22	10	can	can	MD
5767	22	11	form	form	VB
5767	22	12	the	the	DT
5767	22	13	set	set	NN
5767	22	14	of	of	IN
5767	22	15	all	all	PDT
5767	22	16	the	the	DT
5767	22	17	search	search	NN
5767	22	18	key	key	JJ
5767	22	19	values	value	NNS
5767	22	20	that	that	WDT
5767	22	21	will	will	MD
5767	22	22	retrieve	retrieve	VB
5767	22	23	less	less	JJR
5767	22	24	than	than	IN
5767	22	25	n	n	JJ
5767	22	26	items	item	NNS
5767	22	27	.	.	.
5767	23	1	We	-PRON-	PRP
5767	23	2	can	can	MD
5767	23	3	then	then	RB
5767	23	4	determine	determine	VB
5767	23	5	how	how	WRB
5767	23	6	often	often	RB
5767	23	7	search	search	VBP
5767	23	8	key	key	JJ
5767	23	9	values	value	NNS
5767	23	10	from	from	IN
5767	23	11	that	that	DT
5767	23	12	set	set	NN
5767	23	13	are	be	VBP
5767	23	14	requested	request	VBN
5767	23	15	.	.	.
5767	24	1	If	if	IN
5767	24	2	,	,	,
5767	24	3	for	for	IN
5767	24	4	example	example	NN
5767	24	5	,	,	,
5767	24	6	re-	re-	JJ
5767	24	7	quests	quest	NNS
5767	24	8	for	for	IN
5767	24	9	these	these	DT
5767	24	10	values	value	NNS
5767	24	11	occur	occur	VBP
5767	24	12	99	99	CD
5767	24	13	percent	percent	NN
5767	24	14	of	of	IN
5767	24	15	the	the	DT
5767	24	16	time	time	NN
5767	24	17	,	,	,
5767	24	18	then	then	RB
5767	24	19	we	-PRON-	PRP
5767	24	20	can	can	MD
5767	24	21	assert	assert	VB
5767	24	22	that	that	IN
5767	24	23	99	99	CD
5767	24	24	percent	percent	NN
5767	24	25	of	of	IN
5767	24	26	the	the	DT
5767	24	27	time	time	NN
5767	24	28	less	less	JJR
5767	24	29	than	than	IN
5767	24	30	n	n	JJ
5767	24	31	items	item	NNS
5767	24	32	will	will	MD
5767	24	33	be	be	VB
5767	24	34	retrieved	retrieve	VBN
5767	24	35	.	.	.
5767	25	1	If	if	IN
5767	25	2	the	the	DT
5767	25	3	fil	fil	NN
5767	25	4	e	e	NNP
5767	25	5	contains	contain	VBZ
5767	25	6	N	N	NNP
5767	25	7	items	item	NNS
5767	25	8	,	,	,
5767	25	9	then	then	RB
5767	25	10	these	these	DT
5767	25	11	n	n	JJ
5767	25	12	items	item	NNS
5767	25	13	constitute	constitute	VBP
5767	25	14	the	the	DT
5767	25	15	fraction	fraction	NN
5767	25	16	f	f	NN
5767	25	17	=	=	SYM
5767	25	18	~	~	NFP
5767	25	19	of	of	IN
5767	25	20	the	the	DT
5767	25	21	file	file	NN
5767	25	22	.	.	.
5767	26	1	Should	Should	MD
5767	26	2	the	the	DT
5767	26	3	collection	collection	NN
5767	26	4	size	size	NN
5767	26	5	increase	increase	NN
5767	26	6	to	to	IN
5767	26	7	lN	lN	NNP
5767	26	8	,	,	,
5767	26	9	then	then	RB
5767	26	10	the	the	DT
5767	26	11	model	model	NN
5767	26	12	predicts	predict	VBZ
5767	26	13	that	that	IN
5767	26	14	99	99	CD
5767	26	15	percent	percent	NN
5767	26	16	of	of	IN
5767	26	17	the	the	DT
5767	26	18	time	time	NN
5767	26	19	less	less	JJR
5767	26	20	than	than	IN
5767	26	21	f	f	NNP
5767	26	22	(	(	-LRB-
5767	26	23	lN	ln	NN
5767	26	24	)	)	-RRB-
5767	26	25	=	=	NFP
5767	26	26	ln	ln	NN
5767	26	27	items	item	NNS
5767	26	28	would	would	MD
5767	26	29	be	be	VB
5767	26	30	retrieved	retrieve	VBN
5767	26	31	.	.	.
5767	27	1	In	in	IN
5767	27	2	other	other	JJ
5767	27	3	words	word	NNS
5767	27	4	,	,	,
5767	27	5	we	-PRON-	PRP
5767	27	6	have	have	VBP
5767	27	7	precisely	precisely	RB
5767	27	8	the	the	DT
5767	27	9	behavior	behavior	NN
5767	27	10	Kilgour	Kilgour	NNP
5767	27	11	observes	observe	VBZ
5767	27	12	does	do	VBZ
5767	27	13	not	not	RB
5767	27	14	occur	occur	VB
5767	27	15	.	.	.
5767	28	1	This	this	DT
5767	28	2	argument	argument	NN
5767	28	3	shows	show	VBZ
5767	28	4	that	that	IN
5767	28	5	a	a	DT
5767	28	6	simple	simple	JJ
5767	28	7	deterministic	deterministic	JJ
5767	28	8	model	model	NN
5767	28	9	does	do	VBZ
5767	28	10	not	not	RB
5767	28	11	conform	conform	VB
5767	28	12	to	to	TO
5767	28	13	ex-	ex-	DT
5767	28	14	perience	perience	NN
5767	28	15	with	with	IN
5767	28	16	search	search	NN
5767	28	17	keys	key	NNS
5767	28	18	.	.	.
5767	29	1	The	the	DT
5767	29	2	model	model	NN
5767	29	3	breaks	break	VBZ
5767	29	4	down	down	RP
5767	29	5	in	in	IN
5767	29	6	two	two	CD
5767	29	7	ways	way	NNS
5767	29	8	,	,	,
5767	29	9	which	which	WDT
5767	29	10	accounts	account	VBZ
5767	29	11	for	for	IN
5767	29	12	the	the	DT
5767	29	13	dis-	dis-	NN
5767	29	14	crepancy	crepancy	NN
5767	29	15	between	between	IN
5767	29	16	the	the	DT
5767	29	17	results	result	NNS
5767	29	18	derived	derive	VBN
5767	29	19	from	from	IN
5767	29	20	it	-PRON-	PRP
5767	29	21	and	and	CC
5767	29	22	Kilgour	Kilgour	NNP
5767	29	23	's	's	POS
5767	29	24	observations	observation	NNS
5767	29	25	:	:	:
5767	29	26	1	1	LS
5767	29	27	.	.	.
5767	29	28	in	in	IN
5767	29	29	any	any	DT
5767	29	30	actual	actual	JJ
5767	29	31	library	library	NN
5767	29	32	,	,	,
5767	29	33	the	the	DT
5767	29	34	fraction	fraction	NN
5767	29	35	of	of	IN
5767	29	36	the	the	DT
5767	29	37	time	time	NN
5767	29	38	that	that	WDT
5767	29	39	a	a	DT
5767	29	40	particular	particular	JJ
5767	29	41	request	request	NN
5767	29	42	will	will	MD
5767	29	43	appear	appear	VB
5767	29	44	within	within	IN
5767	29	45	a	a	DT
5767	29	46	sequence	sequence	NN
5767	29	47	of	of	IN
5767	29	48	requests	request	NNS
5767	29	49	will	will	MD
5767	29	50	vary	vary	VB
5767	29	51	;	;	:
5767	29	52	and	and	CC
5767	29	53	2	2	CD
5767	29	54	.	.	.
5767	29	55	in	in	IN
5767	29	56	comparing	compare	VBG
5767	29	57	two	two	CD
5767	29	58	different	different	JJ
5767	29	59	samples	sample	NNS
5767	29	60	having	have	VBG
5767	29	61	the	the	DT
5767	29	62	same	same	JJ
5767	29	63	size	size	NN
5767	29	64	,	,	,
5767	29	65	the	the	DT
5767	29	66	number	number	NN
5767	29	67	of	of	IN
5767	29	68	items	item	NNS
5767	29	69	having	have	VBG
5767	29	70	a	a	DT
5767	29	71	given	give	VBN
5767	29	72	search	search	NN
5767	29	73	key	key	JJ
5767	29	74	value	value	NN
5767	29	75	will	will	MD
5767	29	76	vary	vary	VB
5767	29	77	.	.	.
5767	30	1	The	the	DT
5767	30	2	first	first	JJ
5767	30	3	of	of	IN
5767	30	4	these	these	DT
5767	30	5	factors	factor	NNS
5767	30	6	is	be	VBZ
5767	30	7	easily	easily	RB
5767	30	8	dealt	deal	VBN
5767	30	9	with	with	IN
5767	30	10	and	and	CC
5767	30	11	its	-PRON-	PRP$
5767	30	12	analysis	analysis	NN
5767	30	13	will	will	MD
5767	30	14	suggest	suggest	VB
5767	30	15	the	the	DT
5767	30	16	number	number	NN
5767	30	17	of	of	IN
5767	30	18	requests	request	NNS
5767	30	19	to	to	TO
5767	30	20	use	use	VB
5767	30	21	in	in	IN
5767	30	22	a	a	DT
5767	30	23	test	test	NN
5767	30	24	of	of	IN
5767	30	25	search	search	NN
5767	30	26	key	key	JJ
5767	30	27	behavior	behavior	NN
5767	30	28	in	in	IN
5767	30	29	a	a	DT
5767	30	30	given	give	VBN
5767	30	31	library	library	NN
5767	30	32	.	.	.
5767	31	1	For	for	IN
5767	31	2	a	a	DT
5767	31	3	particular	particular	JJ
5767	31	4	collection	collection	NN
5767	31	5	,	,	,
5767	31	6	letS	letS	,
5767	31	7	denote	denote	VB
5767	31	8	the	the	DT
5767	31	9	set	set	NN
5767	31	10	of	of	IN
5767	31	11	search	search	NN
5767	31	12	key	key	JJ
5767	31	13	values	value	NNS
5767	31	14	Statistical	statistical	JJ
5767	31	15	Behavior	Behavior	NNP
5767	31	16	of	of	IN
5767	31	17	Search	Search	NNP
5767	31	18	Keysj	Keysj	NNP
5767	31	19	BOOKSTEIN	BOOKSTEIN	NNP
5767	31	20	111	111	CD
5767	31	21	for	for	IN
5767	31	22	which	which	WDT
5767	31	23	,	,	,
5767	31	24	say	say	VBP
5767	31	25	,	,	,
5767	31	26	twenty	twenty	CD
5767	31	27	or	or	CC
5767	31	28	more	more	JJR
5767	31	29	items	item	NNS
5767	31	30	are	be	VBP
5767	31	31	retrieved	retrieve	VBN
5767	31	32	.	.	.
5767	32	1	We	-PRON-	PRP
5767	32	2	would	would	MD
5767	32	3	like	like	VB
5767	32	4	to	to	TO
5767	32	5	find	find	VB
5767	32	6	the	the	DT
5767	32	7	fraction	fraction	NN
5767	32	8	of	of	IN
5767	32	9	the	the	DT
5767	32	10	time	time	NN
5767	32	11	that	that	WDT
5767	32	12	a	a	DT
5767	32	13	request	request	NN
5767	32	14	in	in	IN
5767	32	15	S	S	NNP
5767	32	16	occurs	occur	VBZ
5767	32	17	in	in	IN
5767	32	18	the	the	DT
5767	32	19	long	long	JJ
5767	32	20	run	run	NN
5767	32	21	;	;	:
5767	32	22	suppose	suppose	VB
5767	32	23	this	this	DT
5767	32	24	value	value	NN
5767	32	25	is	be	VBZ
5767	32	26	in	in	IN
5767	32	27	fact	fact	NN
5767	32	28	q.	q.	.
5767	33	1	Then	then	RB
5767	33	2	among	among	IN
5767	33	3	M	M	NNP
5767	33	4	requests	request	NNS
5767	33	5	,	,	,
5767	33	6	the	the	DT
5767	33	7	probability	probability	NN
5767	33	8	that	that	WDT
5767	33	9	m	m	NNP
5767	33	10	mem-	mem-	NN
5767	33	11	bers	ber	NNS
5767	33	12	of	of	IN
5767	33	13	S	S	NNP
5767	33	14	occur	occur	VBP
5767	33	15	is	be	VBZ
5767	33	16	given	give	VBN
5767	33	17	by	by	IN
5767	33	18	the	the	DT
5767	33	19	binomial	binomial	JJ
5767	33	20	distribution	distribution	NN
5767	33	21	fB(m\q	fB(m\q	''
5767	33	22	,	,	,
5767	33	23	Mi	Mi	NNP
5767	33	24	)	)	-RRB-
5767	33	25	.	.	.
5767	34	1	This	this	DT
5767	34	2	dis-	dis-	IN
5767	34	3	tribution	tribution	NN
5767	34	4	has	have	VBZ
5767	34	5	a	a	DT
5767	34	6	mean	mean	NN
5767	34	7	of	of	IN
5767	34	8	qM	qM	NNP
5767	34	9	and	and	CC
5767	34	10	a	a	DT
5767	34	11	variance	variance	NN
5767	34	12	of	of	IN
5767	34	13	qM(1	qM(1	NNP
5767	34	14	-	-	HYPH
5767	34	15	q	q	NNP
5767	34	16	)	)	-RRB-
5767	34	17	.	.	.
5767	35	1	Should	Should	MD
5767	35	2	we	-PRON-	PRP
5767	35	3	de-	de-	RB
5767	35	4	sire	sire	VBP
5767	35	5	to	to	TO
5767	35	6	estimate	estimate	VB
5767	35	7	the	the	DT
5767	35	8	actual	actual	JJ
5767	35	9	fraction	fraction	NN
5767	35	10	of	of	IN
5767	35	11	the	the	DT
5767	35	12	time	time	NN
5767	35	13	that	that	WDT
5767	35	14	twenty	twenty	CD
5767	35	15	or	or	CC
5767	35	16	more	more	JJR
5767	35	17	items	item	NNS
5767	35	18	will	will	MD
5767	35	19	be	be	VB
5767	35	20	retrieved	retrieve	VBN
5767	35	21	,	,	,
5767	35	22	we	-PRON-	PRP
5767	35	23	can	can	MD
5767	35	24	take	take	VB
5767	35	25	a	a	DT
5767	35	26	sample	sample	NN
5767	35	27	of	of	IN
5767	35	28	M	M	NNP
5767	35	29	requests	request	NNS
5767	35	30	and	and	CC
5767	35	31	compute	compute	NNP
5767	35	32	q	q	NNP
5767	35	33	,	,	,
5767	35	34	the	the	DT
5767	35	35	fraction	fraction	NN
5767	35	36	of	of	IN
5767	35	37	the	the	DT
5767	35	38	requests	request	NNS
5767	35	39	with	with	IN
5767	35	40	search	search	NN
5767	35	41	key	key	JJ
5767	35	42	values	value	NNS
5767	35	43	in	in	IN
5767	35	44	S	S	NNP
5767	35	45	;	;	:
5767	35	46	if	if	IN
5767	35	47	we	-PRON-	PRP
5767	35	48	do	do	VBP
5767	35	49	so	so	RB
5767	35	50	,	,	,
5767	35	51	we	-PRON-	PRP
5767	35	52	will	will	MD
5767	35	53	usually	usually	RB
5767	35	54	get	get	VB
5767	35	55	a	a	DT
5767	35	56	value	value	NN
5767	35	57	for	for	IN
5767	35	58	q	q	NNP
5767	35	59	between	between	IN
5767	35	60	q	q	NNP
5767	35	61	-	-	HYPH
5767	35	62	,	,	,
5767	35	63	/	/	SYM
5767	35	64	M	M	NNP
5767	35	65	v	v	NN
5767	35	66	q	q	NN
5767	35	67	(	(	-LRB-
5767	35	68	1	1	CD
5767	35	69	-	-	HYPH
5767	35	70	q	q	NNP
5767	35	71	)	)	-RRB-
5767	35	72	and	and	CC
5767	35	73	q	q	NN
5767	35	74	+	+	CC
5767	35	75	v2	v2	NN
5767	35	76	M	M	NNP
5767	35	77	v	v	NN
5767	35	78	q	q	NN
5767	35	79	(	(	-LRB-
5767	35	80	1	1	CD
5767	35	81	-	-	HYPH
5767	35	82	q	q	NNP
5767	35	83	)	)	-RRB-
5767	35	84	.	.	.
5767	35	85	'	'	''
5767	36	1	If	if	IN
5767	36	2	for	for	IN
5767	36	3	example	example	NN
5767	36	4	,	,	,
5767	36	5	q	q	NNP
5767	36	6	=	=	SYM
5767	36	7	.01	.01	NFP
5767	36	8	and	and	CC
5767	36	9	M	m	NN
5767	36	10	=	=	SYM
5767	36	11	10,000	10,000	CD
5767	36	12	,	,	,
5767	36	13	we	-PRON-	PRP
5767	36	14	would	would	MD
5767	36	15	tend	tend	VB
5767	36	16	to	to	TO
5767	36	17	find	find	VB
5767	36	18	q	q	NNP
5767	36	19	in	in	IN
5767	36	20	the	the	DT
5767	36	21	interval	interval	NN
5767	36	22	.01	.01	:
5767	36	23	±	±	$
5767	36	24	.002	.002	CD
5767	36	25	.	.	.
5767	37	1	Thus	thus	RB
5767	37	2	the	the	DT
5767	37	3	effect	effect	NN
5767	37	4	of	of	IN
5767	37	5	randomness	randomness	NN
5767	37	6	in	in	IN
5767	37	7	the	the	DT
5767	37	8	ar-	ar-	NN
5767	37	9	rival	rival	NN
5767	37	10	of	of	IN
5767	37	11	requests	request	NNS
5767	37	12	can	can	MD
5767	37	13	easily	easily	RB
5767	37	14	be	be	VB
5767	37	15	controlled	control	VBN
5767	37	16	by	by	IN
5767	37	17	increasing	increase	VBG
5767	37	18	the	the	DT
5767	37	19	number	number	NN
5767	37	20	of	of	IN
5767	37	21	re-	re-	JJ
5767	37	22	quests	quest	NNS
5767	37	23	considered	consider	VBN
5767	37	24	;	;	:
5767	37	25	furthermore	furthermore	RB
5767	37	26	,	,	,
5767	37	27	the	the	DT
5767	37	28	size	size	NN
5767	37	29	of	of	IN
5767	37	30	error	error	NN
5767	37	31	can	can	MD
5767	37	32	be	be	VB
5767	37	33	predicted	predict	VBN
5767	37	34	.	.	.
5767	38	1	We	-PRON-	PRP
5767	38	2	next	next	RB
5767	38	3	introduce	introduce	VBP
5767	38	4	the	the	DT
5767	38	5	second	second	JJ
5767	38	6	factor	factor	NN
5767	38	7	;	;	:
5767	38	8	its	-PRON-	PRP$
5767	38	9	analysis	analysis	NN
5767	38	10	will	will	MD
5767	38	11	suggest	suggest	VB
5767	38	12	how	how	WRB
5767	38	13	the	the	DT
5767	38	14	behavior	behavior	NN
5767	38	15	of	of	IN
5767	38	16	search	search	NN
5767	38	17	keys	key	NNS
5767	38	18	will	will	MD
5767	38	19	change	change	VB
5767	38	20	as	as	IN
5767	38	21	the	the	DT
5767	38	22	collection	collection	NN
5767	38	23	grows	grow	VBZ
5767	38	24	in	in	IN
5767	38	25	size	size	NN
5767	38	26	.	.	.
5767	39	1	For	for	IN
5767	39	2	this	this	DT
5767	39	3	purpose	purpose	NN
5767	39	4	we	-PRON-	PRP
5767	39	5	adopt	adopt	VBP
5767	39	6	a	a	DT
5767	39	7	model	model	NN
5767	39	8	of	of	IN
5767	39	9	collection	collection	NN
5767	39	10	growth	growth	NN
5767	39	11	which	which	WDT
5767	39	12	assumes	assume	VBZ
5767	39	13	that	that	IN
5767	39	14	as	as	IN
5767	39	15	items	item	NNS
5767	39	16	arrive	arrive	VBP
5767	39	17	,	,	,
5767	39	18	they	-PRON-	PRP
5767	39	19	are	be	VBP
5767	39	20	randomly	randomly	RB
5767	39	21	distributed	distribute	VBN
5767	39	22	among	among	IN
5767	39	23	the	the	DT
5767	39	24	search	search	NN
5767	39	25	key	key	JJ
5767	39	26	values	value	NNS
5767	39	27	in	in	IN
5767	39	28	accordance	accordance	NN
5767	39	29	with	with	IN
5767	39	30	some	some	DT
5767	39	31	probability	probability	NN
5767	39	32	distribution	distribution	NN
5767	39	33	.	.	.
5767	40	1	If	if	IN
5767	40	2	we	-PRON-	PRP
5767	40	3	suppose	suppose	VBP
5767	40	4	that	that	IN
5767	40	5	the	the	DT
5767	40	6	probability	probability	NN
5767	40	7	of	of	IN
5767	40	8	an	an	DT
5767	40	9	item	item	NN
5767	40	10	being	be	VBG
5767	40	11	assigned	assign	VBN
5767	40	12	a	a	DT
5767	40	13	specified	specify	VBN
5767	40	14	search	search	NN
5767	40	15	key	key	JJ
5767	40	16	value	value	NN
5767	40	17	,	,	,
5767	40	18	v11	v11	NNP
5767	40	19	is	be	VBZ
5767	40	20	p11	p11	NN
5767	40	21	then	then	RB
5767	40	22	in	in	IN
5767	40	23	a	a	DT
5767	40	24	collection	collection	NN
5767	40	25	of	of	IN
5767	40	26	N	N	NNP
5767	40	27	items	item	NNS
5767	40	28	we	-PRON-	PRP
5767	40	29	may	may	MD
5767	40	30	conclude	conclude	VB
5767	40	31	that	that	IN
5767	40	32	the	the	DT
5767	40	33	probability	probability	NN
5767	40	34	of	of	IN
5767	40	35	n	n	JJ
5767	40	36	items	item	NNS
5767	40	37	having	have	VBG
5767	40	38	that	that	DT
5767	40	39	value	value	NN
5767	40	40	is	be	VBZ
5767	40	41	given	give	VBN
5767	40	42	by	by	IN
5767	40	43	the	the	DT
5767	40	44	binomial	binomial	JJ
5767	40	45	distribution	distribution	NN
5767	40	46	:	:	:
5767	40	47	(	(	-LRB-
5767	40	48	N	n	NN
5767	40	49	)	)	-RRB-
5767	40	50	n	n	NNP
5767	40	51	N	N	NNP
5767	40	52	-	-	HYPH
5767	40	53	n	n	NN
5767	40	54	fu(n	fu(n	CD
5767	40	55	jpbN	jpbN	NNS
5767	40	56	)	)	-RRB-
5767	40	57	=	=	NFP
5767	40	58	7	7	CD
5767	40	59	p1(1-	p1(1-	CD
5767	40	60	p1	p1	NN
5767	40	61	)	)	-RRB-
5767	40	62	•	•	VB
5767	40	63	If	if	IN
5767	40	64	g	g	NN
5767	40	65	'	'	''
5767	40	66	(	(	-LRB-
5767	40	67	v	v	NN
5767	40	68	;	;	:
5767	40	69	)	)	-RRB-
5767	40	70	is	be	VBZ
5767	40	71	the	the	DT
5767	40	72	probability	probability	NN
5767	40	73	that	that	IN
5767	40	74	the	the	DT
5767	40	75	value	value	NN
5767	40	76	v1	v1	NN
5767	40	77	is	be	VBZ
5767	40	78	selected	select	VBN
5767	40	79	from	from	IN
5767	40	80	the	the	DT
5767	40	81	request	request	NN
5767	40	82	population	population	NN
5767	40	83	,	,	,
5767	40	84	then	then	RB
5767	40	85	the	the	DT
5767	40	86	probability	probability	NN
5767	40	87	that	that	IN
5767	40	88	the	the	DT
5767	40	89	"	"	``
5767	40	90	next	next	JJ
5767	40	91	"	"	''
5767	40	92	request	request	NN
5767	40	93	retrieve	retrieve	NN
5767	40	94	n	n	CC
5767	40	95	items	item	NNS
5767	40	96	is	be	VBZ
5767	40	97	given	give	VBN
5767	40	98	by	by	IN
5767	40	99	def	def	NNP
5767	40	100	~~	~~	.
5767	40	101	g'(vt	g'(vt	NNP
5767	40	102	)	)	-RRB-
5767	40	103	fB(njp;,N	fb(njp;,n	NN
5767	40	104	)	)	-RRB-
5767	40	105	=	=	NFP
5767	40	106	fg(p	fg(p	NN
5767	40	107	)	)	-RRB-
5767	40	108	fB(njp	fB(njp	NNS
5767	40	109	,	,	,
5767	40	110	N)dp	N)dp	NNP
5767	40	111	;	;	:
5767	40	112	g(p	g(p	NNP
5767	40	113	)	)	-RRB-
5767	40	114	dp=	dp=	CD
5767	40	115	~	~	NFP
5767	40	116	g'(v	g'(v	ADD
5767	40	117	;	;	:
5767	40	118	)	)	-RRB-
5767	40	119	p	p	LS
5767	40	120	;	;	:
5767	40	121	!	!	.
5767	41	1	P	p	NN
5767	41	2	I	i	NN
5767	41	3	~	~	NFP
5767	41	4	p	p	NN
5767	41	5	+	+	CC
5767	41	6	dp	dp	NN
5767	41	7	is	be	VBZ
5767	41	8	the	the	DT
5767	41	9	probability	probability	NN
5767	41	10	that	that	IN
5767	41	11	a	a	DT
5767	41	12	request	request	NN
5767	41	13	arrive	arrive	VBP
5767	41	14	with	with	IN
5767	41	15	value	value	NN
5767	41	16	p1	p1	NN
5767	41	17	in	in	IN
5767	41	18	the	the	DT
5767	41	19	interval	interval	NN
5767	41	20	(	(	-LRB-
5767	41	21	p	p	NN
5767	41	22	,	,	,
5767	41	23	p	p	NNP
5767	41	24	+	+	CC
5767	41	25	dp	dp	NN
5767	41	26	)	)	-RRB-
5767	41	27	,	,	,
5767	41	28	and	and	CC
5767	41	29	will	will	MD
5767	41	30	be	be	VB
5767	41	31	treated	treat	VBN
5767	41	32	as	as	IN
5767	41	33	a	a	DT
5767	41	34	continuous	continuous	JJ
5767	41	35	function	function	NN
5767	41	36	.	.	.
5767	41	37	"	"	''
5767	41	38	"	"	``
5767	41	39	'	'	''
5767	42	1	Since	since	IN
5767	42	2	the	the	DT
5767	42	3	ex-	ex-	JJ
5767	42	4	pectation	pectation	NN
5767	42	5	of	of	IN
5767	42	6	the	the	DT
5767	42	7	binomial	binomial	JJ
5767	42	8	distribution	distribution	NN
5767	42	9	is	be	VBZ
5767	42	10	given	give	VBN
5767	42	11	by	by	IN
5767	42	12	pN	pN	NNP
5767	42	13	,	,	,
5767	42	14	we	-PRON-	PRP
5767	42	15	have	have	VBP
5767	42	16	de£	de£	NNS
5767	42	17	Nfpg(p)dp	Nfpg(p)dp	NNP
5767	42	18	=	=	SYM
5767	42	19	Np	np	XX
5767	42	20	as	as	IN
5767	42	21	the	the	DT
5767	42	22	expected	expected	JJ
5767	42	23	number	number	NN
5767	42	24	of	of	IN
5767	42	25	items	item	NNS
5767	42	26	retrieved	retrieve	VBN
5767	42	27	by	by	IN
5767	42	28	a	a	DT
5767	42	29	random	random	JJ
5767	42	30	re-	re-	JJ
5767	42	31	quest	quest	NN
5767	42	32	;	;	:
5767	42	33	since	since	IN
5767	42	34	this	this	DT
5767	42	35	is	be	VBZ
5767	42	36	proportional	proportional	JJ
5767	42	37	toN	toN	NNP
5767	42	38	,	,	,
5767	42	39	doubling	double	VBG
5767	42	40	the	the	DT
5767	42	41	size	size	NN
5767	42	42	of	of	IN
5767	42	43	the	the	DT
5767	42	44	collection	collection	NN
5767	42	45	will	will	MD
5767	42	46	,	,	,
5767	42	47	on	on	IN
5767	42	48	the	the	DT
5767	42	49	average	average	NN
5767	42	50	,	,	,
5767	42	51	double	double	PDT
5767	42	52	the	the	DT
5767	42	53	amount	amount	NN
5767	42	54	of	of	IN
5767	42	55	material	material	NN
5767	42	56	ret1·ieved	ret1·ieved	NNP
5767	42	57	.	.	.
5767	43	1	Similarly	similarly	RB
5767	43	2	,	,	,
5767	43	3	the	the	DT
5767	43	4	-	-	HYPH
5767	43	5	2	2	CD
5767	43	6	-	-	HYPH
5767	43	7	2	2	CD
5767	43	8	variance	variance	NN
5767	43	9	,	,	,
5767	43	10	u	u	NNP
5767	43	11	2	2	CD
5767	43	12	,	,	,
5767	43	13	is	be	VBZ
5767	43	14	given	give	VBN
5767	43	15	by	by	IN
5767	43	16	N2	N2	NNP
5767	43	17	(	(	-LRB-
5767	43	18	p	p	NNP
5767	43	19	2	2	CD
5767	43	20	-	-	HYPH
5767	43	21	p	p	NN
5767	43	22	)	)	-RRB-
5767	43	23	+	+	CC
5767	43	24	Nf	Nf	NNP
5767	43	25	p	p	NN
5767	43	26	(	(	-LRB-
5767	43	27	1	1	CD
5767	43	28	-	-	HYPH
5767	43	29	p	p	NN
5767	43	30	)	)	-RRB-
5767	43	31	g	g	NN
5767	43	32	(	(	-LRB-
5767	43	33	p	p	NNP
5767	43	34	)	)	-RRB-
5767	43	35	dp	dp	NNP
5767	43	36	.	.	.
5767	44	1	Should	Should	MD
5767	44	2	p2	p2	NN
5767	44	3	-	-	HYPH
5767	44	4	p	p	NN
5767	44	5	,	,	,
5767	44	6	de£	de£	NNP
5767	44	7	the	the	DT
5767	44	8	variance	variance	NN
5767	44	9	of	of	IN
5767	44	10	p	p	NN
5767	44	11	,	,	,
5767	44	12	be	be	VB
5767	44	13	small	small	JJ
5767	44	14	,	,	,
5767	44	15	this	this	DT
5767	44	16	reduces	reduce	VBZ
5767	44	17	to	to	IN
5767	44	18	Nfp(l	nfp(l	CD
5767	44	19	-	-	HYPH
5767	44	20	p	p	NN
5767	44	21	)	)	-RRB-
5767	44	22	g(p	g(p	NN
5767	44	23	)	)	-RRB-
5767	44	24	dp	dp	NNP
5767	44	25	=	=	FW
5767	44	26	i?N	i?n	UH
5767	44	27	,	,	,
5767	44	28	so	so	IN
5767	44	29	that	that	IN
5767	44	30	approximately	approximately	RB
5767	44	31	95	95	CD
5767	44	32	percent	percent	NN
5767	44	33	of	of	IN
5767	44	34	the	the	DT
5767	44	35	time	time	NN
5767	44	36	the	the	DT
5767	44	37	amount	amount	NN
5767	44	38	of	of	IN
5767	44	39	material	material	NN
5767	44	40	retrieved	retrieve	VBN
5767	44	41	would	would	MD
5767	44	42	be	be	VB
5767	44	43	less	less	JJR
5767	44	44	than	than	IN
5767	44	45	Np	np	NN
5767	44	46	+	+	SYM
5767	44	47	2\1	2\1	CD
5767	44	48	N	n	CD
5767	44	49	a-=	a-=	NN
5767	44	50	N	n	NN
5767	44	51	(	(	-LRB-
5767	44	52	p	p	NN
5767	44	53	+	+	XX
5767	44	54	,	,	,
5767	44	55	~a-	~a-	NFP
5767	44	56	)	)	-RRB-
5767	44	57	.	.	.
5767	45	1	v	v	NNP
5767	45	2	N	N	NNP
5767	45	3	.	.	.
5767	46	1	••	••	VBN
5767	46	2	This	this	DT
5767	46	3	result	result	NN
5767	46	4	would	would	MD
5767	46	5	more	more	RBR
5767	46	6	precisely	precisely	RB
5767	46	7	be	be	VB
5767	46	8	expressed	express	VBN
5767	46	9	as	as	IN
5767	46	10	f	f	LS
5767	46	11	fB(n	fB(n	NNP
5767	46	12	lp	lp	NNP
5767	46	13	,	,	,
5767	46	14	N)dG(p	N)dG(p	NNP
5767	46	15	)	)	-RRB-
5767	46	16	,	,	,
5767	46	17	which	which	WDT
5767	46	18	has	have	VBZ
5767	46	19	the	the	DT
5767	46	20	form	form	NN
5767	46	21	of	of	IN
5767	46	22	a	a	DT
5767	46	23	Stieltjes	Stieltjes	NNP
5767	46	24	integral	integral	JJ
5767	46	25	.	.	.
5767	47	1	The	the	DT
5767	47	2	expression	expression	NN
5767	47	3	used	use	VBN
5767	47	4	in	in	IN
5767	47	5	the	the	DT
5767	47	6	text	text	NN
5767	47	7	is	be	VBZ
5767	47	8	simpler	simple	JJR
5767	47	9	and	and	CC
5767	47	10	reasonably	reasonably	RB
5767	47	11	valid	valid	JJ
5767	47	12	because	because	IN
5767	47	13	of	of	IN
5767	47	14	the	the	DT
5767	47	15	vast	vast	JJ
5767	47	16	number	number	NN
5767	47	17	of	of	IN
5767	47	18	values	value	NNS
5767	47	19	the	the	DT
5767	47	20	search	search	NN
5767	47	21	key	key	NN
5767	47	22	can	can	MD
5767	47	23	take	take	VB
5767	47	24	.	.	.
5767	48	1	I	-PRON-	PRP
5767	48	2	I	-PRON-	PRP
5767	48	3	I	-PRON-	PRP
5767	48	4	J	j	VBP
5767	48	5	I	-PRON-	PRP
5767	48	6	112	112	CD
5767	48	7	Journal	Journal	NNP
5767	48	8	of	of	IN
5767	48	9	Library	Library	NNP
5767	48	10	Automation	Automation	NNP
5767	48	11	Vol	Vol	NNP
5767	48	12	.	.	.
5767	49	1	6/2	6/2	CD
5767	49	2	June	June	NNP
5767	49	3	1973	1973	CD
5767	49	4	It	-PRON-	PRP
5767	49	5	is	be	VBZ
5767	49	6	the	the	DT
5767	49	7	factor	factor	NN
5767	49	8	-	-	:
5767	49	9	+	+	SYM
5767	49	10	2Cf	2cf	CD
5767	49	11	P	p	NN
5767	49	12	vN	vN	NNP
5767	49	13	'	'	''
5767	49	14	and	and	CC
5767	49	15	its	-PRON-	PRP$
5767	49	16	dependence	dependence	NN
5767	49	17	on	on	IN
5767	49	18	N	N	NNP
5767	49	19	,	,	,
5767	49	20	that	that	WDT
5767	49	21	may	may	MD
5767	49	22	account	account	VB
5767	49	23	for	for	IN
5767	49	24	Kilgour	Kilgour	NNP
5767	49	25	's	's	POS
5767	49	26	nonlinearity	nonlinearity	NN
5767	49	27	,	,	,
5767	49	28	and	and	CC
5767	49	29	not	not	RB
5767	49	30	any	any	DT
5767	49	31	property	property	NN
5767	49	32	intrinsic	intrinsic	NN
5767	49	33	in	in	IN
5767	49	34	the	the	DT
5767	49	35	nature	nature	NN
5767	49	36	of	of	IN
5767	49	37	any	any	DT
5767	49	38	type	type	NN
5767	49	39	of	of	IN
5767	49	40	search	search	NN
5767	49	41	key	key	NN
5767	49	42	.	.	.
5767	50	1	Thus	thus	RB
5767	50	2	,	,	,
5767	50	3	to	to	IN
5767	50	4	the	the	DT
5767	50	5	extent	extent	NN
5767	50	6	that	that	IN
5767	50	7	this	this	DT
5767	50	8	model	model	NN
5767	50	9	reflects	reflect	VBZ
5767	50	10	what	what	WP
5767	50	11	is	be	VBZ
5767	50	12	really	really	RB
5767	50	13	happening	happen	VBG
5767	50	14	,	,	,
5767	50	15	the	the	DT
5767	50	16	95	95	CD
5767	50	17	percent	percent	NN
5767	50	18	point	point	NN
5767	50	19	increases	increase	VBZ
5767	50	20	roughly	roughly	RB
5767	50	21	proportionately	proportionately	RB
5767	50	22	with	with	IN
5767	50	23	file	file	NN
5767	50	24	size	size	NN
5767	50	25	;	;	:
5767	50	26	the	the	DT
5767	50	27	"	"	``
5767	50	28	constant	constant	JJ
5767	50	29	"	"	''
5767	50	30	of	of	IN
5767	50	31	proportionality	proportionality	NN
5767	50	32	,	,	,
5767	50	33	however	however	RB
5767	50	34	,	,	,
5767	50	35	is	be	VBZ
5767	50	36	the	the	DT
5767	50	37	sum	sum	NN
5767	50	38	of	of	IN
5767	50	39	two	two	CD
5767	50	40	tem1s	tem1s	RB
5767	50	41	:	:	:
5767	50	42	the	the	DT
5767	50	43	first	first	JJ
5767	50	44	is	be	VBZ
5767	50	45	a	a	DT
5767	50	46	true	true	JJ
5767	50	47	con-	con-	NN
5767	50	48	stant	stant	JJ
5767	50	49	,	,	,
5767	50	50	and	and	CC
5767	50	51	the	the	DT
5767	50	52	second	second	JJ
5767	50	53	is	be	VBZ
5767	50	54	a	a	DT
5767	50	55	term	term	NN
5767	50	56	that	that	WDT
5767	50	57	approaches	approach	VBZ
5767	50	58	zero	zero	CD
5767	50	59	as	as	IN
5767	50	60	the	the	DT
5767	50	61	file	file	NN
5767	50	62	gets	get	VBZ
5767	50	63	larger	large	JJR
5767	50	64	.	.	.
5767	51	1	In	in	IN
5767	51	2	particular	particular	JJ
5767	51	3	,	,	,
5767	51	4	this	this	DT
5767	51	5	model	model	NN
5767	51	6	suggests	suggest	VBZ
5767	51	7	that	that	IN
5767	51	8	we	-PRON-	PRP
5767	51	9	will	will	MD
5767	51	10	never	never	RB
5767	51	11	reach	reach	VB
5767	51	12	a	a	DT
5767	51	13	leveling	leveling	NN
5767	51	14	off	off	RP
5767	51	15	point	point	NN
5767	51	16	-	-	:
5767	51	17	as	as	IN
5767	51	18	the	the	DT
5767	51	19	file	file	NN
5767	51	20	increases	increase	VBZ
5767	51	21	in	in	IN
5767	51	22	size	size	NN
5767	51	23	,	,	,
5767	51	24	the	the	DT
5767	51	25	number	number	NN
5767	51	26	of	of	IN
5767	51	27	items	item	NNS
5767	51	28	retrieved	retrieve	VBN
5767	51	29	will	will	MD
5767	51	30	also	also	RB
5767	51	31	increase	increase	VB
5767	51	32	,	,	,
5767	51	33	and	and	CC
5767	51	34	the	the	DT
5767	51	35	pattern	pattern	NN
5767	51	36	of	of	IN
5767	51	37	increase	increase	NN
5767	51	38	will	will	MD
5767	51	39	become	become	VB
5767	51	40	increasingly	increasingly	RB
5767	51	41	linear	linear	JJ
5767	51	42	.	.	.
5767	52	1	Up	up	IN
5767	52	2	to	to	IN
5767	52	3	this	this	DT
5767	52	4	point	point	NN
5767	52	5	this	this	DT
5767	52	6	discussion	discussion	NN
5767	52	7	has	have	VBZ
5767	52	8	been	be	VBN
5767	52	9	qualitative	qualitative	JJ
5767	52	10	in	in	IN
5767	52	11	nature	nature	NN
5767	52	12	,	,	,
5767	52	13	being	be	VBG
5767	52	14	based	base	VBN
5767	52	15	upon	upon	IN
5767	52	16	general	general	JJ
5767	52	17	statistical	statistical	JJ
5767	52	18	considerations	consideration	NNS
5767	52	19	and	and	CC
5767	52	20	making	make	VBG
5767	52	21	use	use	NN
5767	52	22	of	of	IN
5767	52	23	the	the	DT
5767	52	24	normal	normal	JJ
5767	52	25	approximation	approximation	NN
5767	52	26	to	to	IN
5767	52	27	some	some	DT
5767	52	28	unknown	unknown	JJ
5767	52	29	distribution	distribution	NN
5767	52	30	;	;	:
5767	52	31	its	-PRON-	PRP$
5767	52	32	broad	broad	JJ
5767	52	33	conclusions	conclusion	NNS
5767	52	34	are	be	VBP
5767	52	35	,	,	,
5767	52	36	however	however	RB
5767	52	37	,	,	,
5767	52	38	consistent	consistent	JJ
5767	52	39	with	with	IN
5767	52	40	the	the	DT
5767	52	41	findings	finding	NNS
5767	52	42	of	of	IN
5767	52	43	earlier	early	JJR
5767	52	44	workers	worker	NNS
5767	52	45	and	and	CC
5767	52	46	can	can	MD
5767	52	47	explain	explain	VB
5767	52	48	certai11	certai11	NNS
5767	52	49	unanticipated	unanticipated	JJ
5767	52	50	properties	property	NNS
5767	52	51	of	of	IN
5767	52	52	search	search	NN
5767	52	53	keys	key	NNS
5767	52	54	.	.	.
5767	53	1	To	to	TO
5767	53	2	proceed	proceed	VB
5767	53	3	further	further	JJ
5767	53	4	it	-PRON-	PRP
5767	53	5	will	will	MD
5767	53	6	be	be	VB
5767	53	7	necessary	necessary	JJ
5767	53	8	to	to	TO
5767	53	9	restrict	restrict	VB
5767	53	10	the	the	DT
5767	53	11	form	form	NN
5767	53	12	of	of	IN
5767	53	13	the	the	DT
5767	53	14	function	function	NN
5767	53	15	g(p	g(p	NNP
5767	53	16	)	)	-RRB-
5767	53	17	;	;	:
5767	53	18	tl1is	tl1is	NNP
5767	53	19	will	will	MD
5767	53	20	be	be	VB
5767	53	21	attemped	attempe	VBN
5767	53	22	in	in	IN
5767	53	23	the	the	DT
5767	53	24	following	following	JJ
5767	53	25	section	section	NN
5767	53	26	of	of	IN
5767	53	27	this	this	DT
5767	53	28	paper	paper	NN
5767	53	29	.	.	.
5767	54	1	RELATIONSHIP	RELATIONSHIP	NNS
5767	54	2	OF	of	IN
5767	54	3	MODEL	model	NN
5767	54	4	TO	to	TO
5767	54	5	EARLIER	earlier	RB
5767	54	6	RESEARCH	research	VB
5767	54	7	Interest	interest	NN
5767	54	8	in	in	IN
5767	54	9	access	access	NN
5767	54	10	methods	method	NNS
5767	54	11	that	that	WDT
5767	54	12	are	be	VBP
5767	54	13	appropriate	appropriate	JJ
5767	54	14	for	for	IN
5767	54	15	files	file	NNS
5767	54	16	of	of	IN
5767	54	17	bibliographic	bibliographic	JJ
5767	54	18	data	datum	NNS
5767	54	19	has	have	VBZ
5767	54	20	generated	generate	VBN
5767	54	21	a	a	DT
5767	54	22	considerable	considerable	JJ
5767	54	23	amount	amount	NN
5767	54	24	of	of	IN
5767	54	25	empirical	empirical	JJ
5767	54	26	research	research	NN
5767	54	27	on	on	IN
5767	54	28	search	search	NN
5767	54	29	key	key	JJ
5767	54	30	behavior	behavior	NN
5767	54	31	.	.	.
5767	55	1	Of	of	IN
5767	55	2	necessity	necessity	NN
5767	55	3	,	,	,
5767	55	4	this	this	DT
5767	55	5	pioneering	pioneer	VBG
5767	55	6	work	work	NN
5767	55	7	has	have	VBZ
5767	55	8	been	be	VBN
5767	55	9	of	of	IN
5767	55	10	a	a	DT
5767	55	11	descriptive	descriptive	JJ
5767	55	12	nature	nature	NN
5767	55	13	,	,	,
5767	55	14	resulting	result	VBG
5767	55	15	in	in	IN
5767	55	16	data	datum	NNS
5767	55	17	showing	show	VBG
5767	55	18	search	search	NN
5767	55	19	key	key	JJ
5767	55	20	behavior	behavior	NN
5767	55	21	in	in	IN
5767	55	22	specific	specific	JJ
5767	55	23	environ-	environ-	NN
5767	55	24	ments	ment	NNS
5767	55	25	.	.	.
5767	56	1	While	while	IN
5767	56	2	these	these	DT
5767	56	3	efforts	effort	NNS
5767	56	4	have	have	VBP
5767	56	5	lent	lend	VBN
5767	56	6	a	a	DT
5767	56	7	good	good	JJ
5767	56	8	deal	deal	NN
5767	56	9	of	of	IN
5767	56	10	insight	insight	NN
5767	56	11	into	into	IN
5767	56	12	the	the	DT
5767	56	13	nature	nature	NN
5767	56	14	of	of	IN
5767	56	15	search	search	NN
5767	56	16	keys	key	NNS
5767	56	17	,	,	,
5767	56	18	the	the	DT
5767	56	19	basic	basic	JJ
5767	56	20	weakness	weakness	NN
5767	56	21	of	of	IN
5767	56	22	such	such	JJ
5767	56	23	research	research	NN
5767	56	24	lies	lie	VBZ
5767	56	25	in	in	IN
5767	56	26	the	the	DT
5767	56	27	difficulty	difficulty	NN
5767	56	28	of	of	IN
5767	56	29	extending	extend	VBG
5767	56	30	these	these	DT
5767	56	31	findings	finding	NNS
5767	56	32	to	to	IN
5767	56	33	other	other	JJ
5767	56	34	situations	situation	NNS
5767	56	35	.	.	.
5767	57	1	One	one	CD
5767	57	2	purpose	purpose	NN
5767	57	3	of	of	IN
5767	57	4	a	a	DT
5767	57	5	mathematical	mathematical	JJ
5767	57	6	model	model	NN
5767	57	7	such	such	JJ
5767	57	8	·	·	NFP
5767	57	9	as	as	IN
5767	57	10	.	.	.
5767	58	1	the	the	DT
5767	58	2	one	one	NN
5767	58	3	being	be	VBG
5767	58	4	developed	develop	VBN
5767	58	5	here	here	RB
5767	58	6	is	be	VBZ
5767	58	7	to	to	TO
5767	58	8	provide	provide	VB
5767	58	9	this	this	DT
5767	58	10	increased	increase	VBN
5767	58	11	generality	generality	NN
5767	58	12	by	by	IN
5767	58	13	representing	represent	VBG
5767	58	14	in	in	IN
5767	58	15	a	a	DT
5767	58	16	concise	concise	NN
5767	58	17	and	and	CC
5767	58	18	easily	easily	RB
5767	58	19	manipulated	manipulate	VBN
5767	58	20	form	form	NN
5767	58	21	the	the	DT
5767	58	22	results	result	NNS
5767	58	23	of	of	IN
5767	58	24	previous	previous	JJ
5767	58	25	research	research	NN
5767	58	26	.	.	.
5767	59	1	It	-PRON-	PRP
5767	59	2	is	be	VBZ
5767	59	3	accordingly	accordingly	RB
5767	59	4	of	of	IN
5767	59	5	interest	interest	NN
5767	59	6	to	to	TO
5767	59	7	indicate	indicate	VB
5767	59	8	the	the	DT
5767	59	9	re-	re-	JJ
5767	59	10	lationship	lationship	NN
5767	59	11	between	between	IN
5767	59	12	previous	previous	JJ
5767	59	13	work	work	NN
5767	59	14	on	on	IN
5767	59	15	search	search	NN
5767	59	16	keys	key	NNS
5767	59	17	and	and	CC
5767	59	18	our	-PRON-	PRP$
5767	59	19	model	model	NN
5767	59	20	.	.	.
5767	60	1	Research	research	NN
5767	60	2	on	on	IN
5767	60	3	search	search	NN
5767	60	4	key	key	JJ
5767	60	5	performance	performance	NN
5767	60	6	has	have	VBZ
5767	60	7	been	be	VBN
5767	60	8	of	of	IN
5767	60	9	two	two	CD
5767	60	10	kinds	kind	NNS
5767	60	11	.	.	.
5767	61	1	The	the	DT
5767	61	2	fi.rst	fi.rst	``
5767	61	3	kind	kind	NN
5767	61	4	seeks	seek	VBZ
5767	61	5	.to	.to	.
5767	61	6	answer	answer	VB
5767	61	7	the	the	DT
5767	61	8	question	question	NN
5767	61	9	:	:	:
5767	61	10	for	for	IN
5767	61	11	any	any	DT
5767	61	12	number	number	NN
5767	61	13	,	,	,
5767	61	14	n	n	CC
5767	61	15	,	,	,
5767	61	16	how	how	WRB
5767	61	17	many	many	JJ
5767	61	18	search	search	NN
5767	61	19	key	key	JJ
5767	61	20	values	value	NNS
5767	61	21	retrieve	retrieve	VBP
5767	61	22	n	n	DT
5767	61	23	items	item	NNS
5767	61	24	?	?	.
5767	62	1	The	the	DT
5767	62	2	answer	answer	NN
5767	62	3	to	to	IN
5767	62	4	this	this	DT
5767	62	5	question	question	NN
5767	62	6	depends	depend	VBZ
5767	62	7	only	only	RB
5767	62	8	on	on	IN
5767	62	9	the	the	DT
5767	62	10	search	search	NN
5767	62	11	key	key	NN
5767	62	12	and	and	CC
5767	62	13	the	the	DT
5767	62	14	collection	collection	NN
5767	62	15	;	;	:
5767	62	16	it	-PRON-	PRP
5767	62	17	is	be	VBZ
5767	62	18	independent	independent	JJ
5767	62	19	of	of	IN
5767	62	20	the	the	DT
5767	62	21	pattern	pattern	NN
5767	62	22	of	of	IN
5767	62	23	re-	re-	JJ
5767	62	24	quest	quest	NN
5767	62	25	arrivals	arrival	NNS
5767	62	26	.	.	.
5767	63	1	The	the	DT
5767	63	2	second	second	JJ
5767	63	3	kind	kind	NN
5767	63	4	of	of	IN
5767	63	5	research	research	NN
5767	63	6	involves	involve	VBZ
5767	63	7	the	the	DT
5767	63	8	·	·	NFP
5767	63	9	actual	actual	JJ
5767	63	10	arrival	arrival	NN
5767	63	11	of	of	IN
5767	63	12	requests	request	NNS
5767	63	13	;	;	:
5767	63	14	it	-PRON-	PRP
5767	63	15	tries	try	VBZ
5767	63	16	to	to	TO
5767	63	17	answer	answer	VB
5767	63	18	the	the	DT
5767	63	19	question	question	NN
5767	63	20	:	:	:
5767	63	21	for	for	IN
5767	63	22	any	any	DT
5767	63	23	number	number	NN
5767	63	24	n	n	JJ
5767	63	25	,	,	,
5767	63	26	how	how	WRB
5767	63	27	frequently	frequently	RB
5767	63	28	will	will	MD
5767	63	29	requests	request	NNS
5767	63	30	resulting	result	VBG
5767	63	31	in	in	IN
5767	63	32	the	the	DT
5767	63	33	retrieval	retrieval	NN
5767	63	34	of	of	IN
5767	63	35	n	n	JJ
5767	63	36	items	item	NNS
5767	63	37	occur	occur	VBP
5767	63	38	?	?	.
5767	64	1	To	to	TO
5767	64	2	discuss	discuss	VB
5767	64	3	this	this	DT
5767	64	4	research	research	NN
5767	64	5	in	in	IN
5767	64	6	terms	term	NNS
5767	64	7	of	of	IN
5767	64	8	our	-PRON-	PRP$
5767	64	9	model	model	NN
5767	64	10	requires	require	VBZ
5767	64	11	a	a	DT
5767	64	12	closer	close	JJR
5767	64	13	examina-	examina-	NN
5767	64	14	def	def	JJ
5767	64	15	tion	tion	NN
5767	64	16	of	of	IN
5767	64	17	the	the	DT
5767	64	18	function	function	NN
5767	64	19	g	g	NN
5767	64	20	(	(	-LRB-
5767	64	21	p	p	NN
5767	64	22	)	)	-RRB-
5767	64	23	previously	previously	RB
5767	64	24	defined	define	VBN
5767	64	25	.	.	.
5767	65	1	We	-PRON-	PRP
5767	65	2	recall	recall	VBP
5767	65	3	that	that	IN
5767	65	4	g	g	NN
5767	65	5	(	(	-LRB-
5767	65	6	p	p	NNP
5767	65	7	)	)	-RRB-
5767	65	8	dp	dp	NNP
5767	65	9	=	=	SYM
5767	65	10	=	=	SYM
5767	65	11	~	~	NFP
5767	65	12	g'(v1	g'(v1	NNP
5767	65	13	)	)	-RRB-
5767	65	14	,	,	,
5767	65	15	with	with	IN
5767	65	16	dp	dp	NNP
5767	65	17	being	be	VBG
5767	65	18	a	a	DT
5767	65	19	small	small	JJ
5767	65	20	number	number	NN
5767	65	21	.	.	.
5767	66	1	Thus	thus	RB
5767	66	2	g(p	g(p	NNP
5767	66	3	)	)	-RRB-
5767	66	4	is	be	VBZ
5767	66	5	determined	determine	VBN
5767	66	6	P	p	NN
5767	66	7	~	~	NFP
5767	66	8	PI	pi	NN
5767	66	9	~	~	NFP
5767	66	10	p+dp	p+dp	:
5767	66	11	by	by	IN
5767	66	12	two	two	CD
5767	66	13	factors	factor	NNS
5767	66	14	:	:	:
5767	66	15	Statistical	statistical	JJ
5767	66	16	Behavior	Behavior	NNP
5767	66	17	of	of	IN
5767	66	18	Semch	Semch	NNP
5767	66	19	KeysjBOOKSTEIN	KeysjBOOKSTEIN	NNP
5767	66	20	113	113	CD
5767	66	21	a.	a.	NN
5767	67	1	The	the	DT
5767	67	2	number	number	NN
5767	67	3	of	of	IN
5767	67	4	search	search	NN
5767	67	5	key	key	JJ
5767	67	6	values	value	NNS
5767	67	7	in	in	IN
5767	67	8	the	the	DT
5767	67	9	interval	interval	NN
5767	67	10	(	(	-LRB-
5767	67	11	p	p	NN
5767	67	12	,	,	,
5767	67	13	p	p	NNP
5767	67	14	+	+	CC
5767	67	15	dp	dp	NN
5767	67	16	)	)	-RRB-
5767	67	17	.	.	.
5767	68	1	Let	let	VB
5767	68	2	us	-PRON-	PRP
5767	68	3	denote	denote	VB
5767	68	4	this	this	DT
5767	68	5	value	value	NN
5767	68	6	by	by	IN
5767	68	7	f(p	f(p	NNP
5767	68	8	)	)	-RRB-
5767	68	9	dp	dp	NNP
5767	68	10	,	,	,
5767	68	11	so	so	RB
5767	68	12	f(p	f(p	NNP
5767	68	13	)	)	-RRB-
5767	68	14	is	be	VBZ
5767	68	15	the	the	DT
5767	68	16	density	density	NN
5767	68	17	of	of	IN
5767	68	18	search	search	NN
5767	68	19	keys	key	NNS
5767	68	20	at	at	IN
5767	68	21	p.	p.	NN
5767	68	22	We	-PRON-	PRP
5767	68	23	make	make	VBP
5767	68	24	use	use	NN
5767	68	25	here	here	RB
5767	68	26	of	of	IN
5767	68	27	the	the	DT
5767	68	28	fact	fact	NN
5767	68	29	that	that	IN
5767	68	30	although	although	IN
5767	68	31	the	the	DT
5767	68	32	number	number	NN
5767	68	33	of	of	IN
5767	68	34	possible	possible	JJ
5767	68	35	search	search	NN
5767	68	36	key	key	JJ
5767	68	37	values	value	NNS
5767	68	38	is	be	VBZ
5767	68	39	finite	finite	JJ
5767	68	40	,	,	,
5767	68	41	the	the	DT
5767	68	42	number	number	NN
5767	68	43	is	be	VBZ
5767	68	44	very	very	RB
5767	68	45	large	large	JJ
5767	68	46	,	,	,
5767	68	47	so	so	IN
5767	68	48	their	-PRON-	PRP$
5767	68	49	.	.	.
5767	69	1	distribu-	distribu-	JJ
5767	69	2	tion	tion	NN
5767	69	3	can	can	MD
5767	69	4	be	be	VB
5767	69	5	thought	think	VBN
5767	69	6	of	of	IN
5767	69	7	as	as	RB
5767	69	8	continuous	continuous	JJ
5767	69	9	.	.	.
5767	70	1	b.	b.	NNP
5767	71	1	The	the	DT
5767	71	2	average	average	JJ
5767	71	3	probability	probability	NN
5767	71	4	of	of	IN
5767	71	5	search	search	NN
5767	71	6	keys	key	NNS
5767	71	7	,	,	,
5767	71	8	with	with	IN
5767	71	9	values	value	NNS
5767	71	10	p	p	NNP
5767	71	11	1	1	CD
5767	71	12	near	near	IN
5767	71	13	p	p	NN
5767	71	14	,	,	,
5767	71	15	being	be	VBG
5767	71	16	requested	request	VBN
5767	71	17	.	.	.
5767	72	1	We	-PRON-	PRP
5767	72	2	shall	shall	MD
5767	72	3	refer	refer	VB
5767	72	4	to	to	IN
5767	72	5	this	this	DT
5767	72	6	quantity	quantity	NN
5767	72	7	as	as	IN
5767	72	8	g"(p	g"(p	NNP
5767	72	9	)	)	-RRB-
5767	72	10	.	.	.
5767	73	1	By	by	IN
5767	73	2	combining	combine	VBG
5767	73	3	these	these	DT
5767	73	4	factors	factor	NNS
5767	73	5	we	-PRON-	PRP
5767	73	6	have	have	VBP
5767	73	7	g(p	g(p	NNP
5767	73	8	)	)	-RRB-
5767	73	9	=	=	NFP
5767	73	10	g"(p	g"(p	FW
5767	73	11	)	)	-RRB-
5767	73	12	f(p	f(p	NNP
5767	73	13	)	)	-RRB-
5767	73	14	.	.	.
5767	74	1	·	·	NFP
5767	74	2	In	in	IN
5767	74	3	terms	term	NNS
5767	74	4	of	of	IN
5767	74	5	this	this	DT
5767	74	6	discussion	discussion	NN
5767	74	7	,	,	,
5767	74	8	the	the	DT
5767	74	9	first	first	JJ
5767	74	10	type	type	NN
5767	74	11	of	of	IN
5767	74	12	research	research	NN
5767	74	13	described	describe	VBN
5767	74	14	above	above	RB
5767	74	15	.	.	.
5767	75	1	is	be	VBZ
5767	75	2	in	in	IN
5767	75	3	fact	fact	NN
5767	75	4	estimating	estimate	VBG
5767	75	5	f(p	f(p	CD
5767	75	6	)	)	-RRB-
5767	75	7	:	:	:
5767	75	8	if	if	IN
5767	75	9	there	there	EX
5767	75	10	ares	are	VBZ
5767	75	11	search	search	NN
5767	75	12	key	key	JJ
5767	75	13	values	value	NNS
5767	75	14	that	that	WDT
5767	75	15	retrieve	retrieve	VBP
5767	75	16	n	n	DT
5767	75	17	items	item	NNS
5767	75	18	from	from	IN
5767	75	19	a	a	DT
5767	75	20	collection	collection	NN
5767	75	21	of	of	IN
5767	75	22	N	N	NNP
5767	75	23	items	item	NNS
5767	75	24	,	,	,
5767	75	25	then	then	RB
5767	75	26	sis	sis	VB
5767	75	27	an	an	DT
5767	75	28	estimate	estimate	NN
5767	75	29	of	of	IN
5767	75	30	this	this	DT
5767	75	31	relation	relation	NN
5767	75	32	uses	use	VBZ
5767	75	33	_	_	NNP
5767	75	34	!	!	.
5767	75	35	_	_	NNP
5767	75	36	f	f	NNP
5767	75	37	(	(	-LRB-
5767	75	38	~	~	NFP
5767	75	39	)	)	-RRB-
5767	75	40	·	·	NFP
5767	75	41	N	n	NN
5767	75	42	N	n	NN
5767	75	43	'	'	CC
5767	75	44	n	n	CC
5767	75	45	+	+	CC
5767	75	46	~~	~~	NNS
5767	75	47	n	n	NN
5767	75	48	=	=	SYM
5767	75	49	pN	pN	NNP
5767	75	50	,	,	,
5767	75	51	and	and	CC
5767	75	52	dp	dp	NNP
5767	75	53	=	=	SYM
5767	75	54	N	n	NN
5767	75	55	n-	n-	XX
5767	75	56	~	~	NFP
5767	75	57	1	1	CD
5767	75	58	N	n	NN
5767	75	59	N	n	CC
5767	75	60	'	'	''
5767	75	61	The	the	DT
5767	75	62	second	second	JJ
5767	75	63	kind	kind	NN
5767	75	64	of	of	IN
5767	75	65	research	research	NN
5767	75	66	directly	directly	RB
5767	75	67	estimates	estimate	VBZ
5767	75	68	g	g	NNP
5767	75	69	(	(	-LRB-
5767	75	70	p	p	NNP
5767	75	71	)	)	-RRB-
5767	75	72	.	.	.
5767	76	1	Guthrie	Guthrie	NNP
5767	76	2	,	,	,
5767	76	3	in	in	IN
5767	76	4	a	a	DT
5767	76	5	recent	recent	JJ
5767	76	6	paper	paper	NN
5767	76	7	,	,	,
5767	76	8	provides	provide	VBZ
5767	76	9	a	a	DT
5767	76	10	bridge	bridge	NN
5767	76	11	between	between	IN
5767	76	12	the	the	DT
5767	76	13	two	two	CD
5767	76	14	types	type	NNS
5767	76	15	of	of	IN
5767	76	16	research	research	NN
5767	76	17	by	by	IN
5767	76	18	discussing	discuss	VBG
5767	76	19	his	-PRON-	PRP$
5767	76	20	findings	finding	NNS
5767	76	21	in	in	IN
5767	76	22	terms	term	NNS
5767	76	23	of	of	IN
5767	76	24	two	two	CD
5767	76	25	models.6	models.6	CD
5767	76	26	One	one	CD
5767	76	27	of	of	IN
5767	76	28	his	-PRON-	PRP$
5767	76	29	models	model	NNS
5767	76	30	,	,	,
5767	76	31	which	which	WDT
5767	76	32	asserts	assert	VBZ
5767	76	33	that	that	IN
5767	76	34	each	each	DT
5767	76	35	search	search	NN
5767	76	36	key	key	JJ
5767	76	37	value	value	NN
5767	76	38	has	have	VBZ
5767	76	39	an	an	DT
5767	76	40	equal	equal	JJ
5767	76	41	chance	chance	NN
5767	76	42	of	of	IN
5767	76	43	being	be	VBG
5767	76	44	requested	request	VBN
5767	76	45	,	,	,
5767	76	46	is	be	VBZ
5767	76	47	equivalent	equivalent	JJ
5767	76	48	to	to	IN
5767	76	49	the	the	DT
5767	76	50	assumption	assumption	NN
5767	76	51	that	that	IN
5767	76	52	g"(p	g"(p	NNP
5767	76	53	)	)	-RRB-
5767	76	54	=	=	SYM
5767	76	55	1	1	CD
5767	76	56	,	,	,
5767	76	57	and	and	CC
5767	76	58	g(p	g(p	NNP
5767	76	59	)	)	-RRB-
5767	76	60	=	=	NFP
5767	76	61	f(p	f(p	CD
5767	76	62	)	)	-RRB-
5767	76	63	.	.	.
5767	77	1	Guthrie	Guthrie	NNP
5767	77	2	finds	find	VBZ
5767	77	3	that	that	IN
5767	77	4	this	this	DT
5767	77	5	is	be	VBZ
5767	77	6	not	not	RB
5767	77	7	an	an	DT
5767	77	8	adequate	adequate	JJ
5767	77	9	representation	representation	NN
5767	77	10	of	of	IN
5767	77	11	his	-PRON-	PRP$
5767	77	12	data	datum	NNS
5767	77	13	.	.	.
5767	78	1	Guthrie	Guthrie	NNP
5767	78	2	's	's	POS
5767	78	3	second	second	JJ
5767	78	4	model	model	NN
5767	78	5	asserts	assert	VBZ
5767	78	6	that	that	IN
5767	78	7	each	each	DT
5767	78	8	item	item	NN
5767	78	9	has	have	VBZ
5767	78	10	an	an	DT
5767	78	11	equal	equal	JJ
5767	78	12	chance	chance	NN
5767	78	13	of	of	IN
5767	78	14	being	be	VBG
5767	78	15	requested	request	VBN
5767	78	16	.	.	.
5767	79	1	In	in	IN
5767	79	2	our	-PRON-	PRP$
5767	79	3	terms	term	NNS
5767	79	4	this	this	DT
5767	79	5	becomes	become	VBZ
5767	79	6	g	g	NNP
5767	79	7	'	'	''
5767	79	8	(	(	-LRB-
5767	79	9	p	p	NN
5767	79	10	)	)	-RRB-
5767	79	11	ap	ap	UH
5767	79	12	,	,	,
5767	79	13	and	and	CC
5767	79	14	g	g	NNP
5767	79	15	(	(	-LRB-
5767	79	16	p	p	NNP
5767	79	17	)	)	-RRB-
5767	79	18	apf	apf	NNP
5767	79	19	(	(	-LRB-
5767	79	20	p	p	NNP
5767	79	21	)	)	-RRB-
5767	79	22	.	.	.
5767	80	1	This	this	DT
5767	80	2	model	model	NN
5767	80	3	,	,	,
5767	80	4	while	while	IN
5767	80	5	an	an	DT
5767	80	6	improvement	improvement	NN
5767	80	7	over	over	IN
5767	80	8	the	the	DT
5767	80	9	first	first	JJ
5767	80	10	,	,	,
5767	80	11	still	still	RB
5767	80	12	disagrees	disagree	VBZ
5767	80	13	with	with	IN
5767	80	14	the	the	DT
5767	80	15	data	datum	NNS
5767	80	16	.	.	.
5767	81	1	Furthermore	furthermore	RB
5767	81	2	,	,	,
5767	81	3	these	these	DT
5767	81	4	models	model	NNS
5767	81	5	do	do	VBP
5767	81	6	not	not	RB
5767	81	7	estimate	estimate	VB
5767	81	8	f	f	NNP
5767	81	9	(	(	-LRB-
5767	81	10	p	p	NNP
5767	81	11	)	)	-RRB-
5767	81	12	;	;	:
5767	81	13	even	even	RB
5767	81	14	if	if	IN
5767	81	15	Guthrie	Guthrie	NNP
5767	81	16	's	's	POS
5767	81	17	model	model	NN
5767	81	18	were	be	VBD
5767	81	19	correct	correct	JJ
5767	81	20	,	,	,
5767	81	21	we	-PRON-	PRP
5767	81	22	would	would	MD
5767	81	23	not	not	RB
5767	81	24	know	know	VB
5767	81	25	the	the	DT
5767	81	26	probability	probability	NN
5767	81	27	that	that	IN
5767	81	28	n	n	DT
5767	81	29	items	item	NNS
5767	81	30	would	would	MD
5767	81	31	be	be	VB
5767	81	32	re	re	VBN
5767	81	33	-	-	VBN
5767	81	34	trieved	trieve	VBN
5767	81	35	until	until	IN
5767	81	36	we	-PRON-	PRP
5767	81	37	were	be	VBD
5767	81	38	told	tell	VBN
5767	81	39	how	how	WRB
5767	81	40	many	many	JJ
5767	81	41	search	search	NN
5767	81	42	key	key	JJ
5767	81	43	values	value	NNS
5767	81	44	contained	contain	VBD
5767	81	45	n	n	JJ
5767	81	46	items	item	NNS
5767	81	47	.	.	.
5767	82	1	In	in	IN
5767	82	2	the	the	DT
5767	82	3	next	next	JJ
5767	82	4	section	section	NN
5767	82	5	we	-PRON-	PRP
5767	82	6	will	will	MD
5767	82	7	try	try	VB
5767	82	8	to	to	TO
5767	82	9	remedy	remedy	VB
5767	82	10	this	this	DT
5767	82	11	situation	situation	NN
5767	82	12	by	by	IN
5767	82	13	means	mean	NNS
5767	82	14	of	of	IN
5767	82	15	a	a	DT
5767	82	16	two	two	CD
5767	82	17	paramete	paramete	NN
5767	82	18	r	r	NN
5767	82	19	representation	representation	NN
5767	82	20	of	of	IN
5767	82	21	g	g	NNP
5767	82	22	(	(	-LRB-
5767	82	23	p	p	NNP
5767	82	24	)	)	-RRB-
5767	82	25	.	.	.
5767	83	1	A	a	DT
5767	83	2	REPRESENTATION	representation	NN
5767	83	3	OF	of	IN
5767	83	4	f(p	f(p	NN
5767	83	5	)	)	-RRB-
5767	83	6	To	to	TO
5767	83	7	get	get	VB
5767	83	8	a	a	DT
5767	83	9	more	more	RBR
5767	83	10	detailed	detailed	JJ
5767	83	11	account	account	NN
5767	83	12	of	of	IN
5767	83	13	search	search	NN
5767	83	14	key	key	JJ
5767	83	15	behavior	behavior	NN
5767	83	16	by	by	IN
5767	83	17	experiment	experiment	NN
5767	83	18	is	be	VBZ
5767	83	19	difficult	difficult	JJ
5767	83	20	since	since	IN
5767	83	21	the	the	DT
5767	83	22	two	two	CD
5767	83	23	aspects	aspect	NNS
5767	83	24	of	of	IN
5767	83	25	randomness	randomness	NN
5767	83	26	already	already	RB
5767	83	27	discussed	discuss	VBN
5767	83	28	are	be	VBP
5767	83	29	con-	con-	NN
5767	83	30	founded	found	VBN
5767	83	31	;	;	:
5767	83	32	the	the	DT
5767	83	33	experimenter	experimenter	NN
5767	83	34	only	only	RB
5767	83	35	sees	see	VBZ
5767	83	36	the	the	DT
5767	83	37	combined	combine	VBN
5767	83	38	effect	effect	NN
5767	83	39	.	.	.
5767	84	1	We	-PRON-	PRP
5767	84	2	will	will	MD
5767	84	3	,	,	,
5767	84	4	however	however	RB
5767	84	5	,	,	,
5767	84	6	try	try	VB
5767	84	7	to	to	TO
5767	84	8	estimate	estimate	VB
5767	84	9	the	the	DT
5767	84	10	distribution	distribution	NN
5767	84	11	g	g	NN
5767	84	12	(	(	-LRB-
5767	84	13	p	p	NNP
5767	84	14	)	)	-RRB-
5767	84	15	by	by	IN
5767	84	16	a	a	DT
5767	84	17	distribution	distribution	NN
5767	84	18	of	of	IN
5767	84	19	the	the	DT
5767	84	20	form	form	NN
5767	84	21	(	(	-LRB-
5767	84	22	a	a	DT
5767	84	23	+	+	SYM
5767	84	24	{	{	-LRB-
5767	84	25	3	3	CD
5767	84	26	+	+	SYM
5767	84	27	1	1	CD
5767	84	28	)	)	-RRB-
5767	84	29	!	!	.
5767	85	1	a	a	DT
5767	85	2	(	(	-LRB-
5767	85	3	1	1	CD
5767	85	4	-	-	HYPH
5767	85	5	)	)	-RRB-
5767	85	6	f3	f3	NN
5767	85	7	a!f3	a!f3	NN
5767	85	8	!	!	.
5767	86	1	P	p	NN
5767	86	2	p.	p.	NN
5767	86	3	We	-PRON-	PRP
5767	86	4	believe	believe	VBP
5767	86	5	that	that	IN
5767	86	6	such	such	PDT
5767	86	7	an	an	DT
5767	86	8	attempt	attempt	NN
5767	86	9	is	be	VBZ
5767	86	10	reasonable	reasonable	JJ
5767	86	11	on	on	IN
5767	86	12	three	three	CD
5767	86	13	grounds	ground	NNS
5767	86	14	:	:	:
5767	86	15	a.	a.	NN
5767	87	1	It	-PRON-	PRP
5767	87	2	is	be	VBZ
5767	87	3	not	not	RB
5767	87	4	possible	possible	JJ
5767	87	5	to	to	TO
5767	87	6	find	find	VB
5767	87	7	g(p	g(p	NNP
5767	87	8	)	)	-RRB-
5767	87	9	exactly	exactly	RB
5767	87	10	,	,	,
5767	87	11	and	and	CC
5767	87	12	moreover	moreover	RB
5767	87	13	,	,	,
5767	87	14	it	-PRON-	PRP
5767	87	15	is	be	VBZ
5767	87	16	not	not	RB
5767	87	17	clear	clear	JJ
5767	87	18	that	that	IN
5767	87	19	this	this	DT
5767	87	20	would	would	MD
5767	87	21	be	be	VB
5767	87	22	desirable	desirable	JJ
5767	87	23	.	.	.
5767	88	1	We	-PRON-	PRP
5767	88	2	are	be	VBP
5767	88	3	interested	interested	JJ
5767	88	4	in	in	IN
5767	88	5	a	a	DT
5767	88	6	reasonable	reasonable	JJ
5767	88	7	ap-	ap-	NN
5767	88	8	proximation	proximation	NN
5767	88	9	that	that	WDT
5767	88	10	is	be	VBZ
5767	88	11	satisfactory	satisfactory	JJ
5767	88	12	for	for	IN
5767	88	13	decision	decision	NN
5767	88	14	-	-	HYPH
5767	88	15	making	make	VBG
5767	88	16	purposes	purpose	NNS
5767	88	17	;	;	:
5767	88	18	b.	b.	NNP
5767	89	1	The	the	DT
5767	89	2	above	above	JJ
5767	89	3	distribution	distribution	NN
5767	89	4	assumes	assume	VBZ
5767	89	5	a	a	DT
5767	89	6	wide	wide	JJ
5767	89	7	variety	variety	NN
5767	89	8	of	of	IN
5767	89	9	shapes	shape	NNS
5767	89	10	as	as	IN
5767	89	11	a	a	DT
5767	89	12	and	and	CC
5767	89	13	f3	f3	NNP
5767	89	14	vary	vary	VBP
5767	89	15	;	;	:
5767	89	16	it	-PRON-	PRP
5767	89	17	seems	seem	VBZ
5767	89	18	likely	likely	JJ
5767	89	19	that	that	IN
5767	89	20	values	value	NNS
5767	89	21	of	of	IN
5767	89	22	a	a	DT
5767	89	23	and	and	CC
5767	89	24	f3	f3	NN
5767	89	25	can	can	MD
5767	89	26	be	be	VB
5767	89	27	found	find	VBN
5767	89	28	for	for	IN
5767	89	29	which	which	WDT
5767	89	30	114	114	CD
5767	89	31	Journal	Journal	NNP
5767	89	32	of	of	IN
5767	89	33	Library	Library	NNP
5767	89	34	Automation	Automation	NNP
5767	89	35	Vol	Vol	NNP
5767	89	36	.	.	.
5767	90	1	6/	6/	CD
5767	90	2	2	2	CD
5767	90	3	June	June	NNP
5767	90	4	1973	1973	CD
5767	90	5	this	this	DT
5767	90	6	distribution	distribution	NN
5767	90	7	is	be	VBZ
5767	90	8	close	close	JJ
5767	90	9	enough	enough	RB
5767	90	10	to	to	IN
5767	90	11	g	g	NNP
5767	90	12	(	(	-LRB-
5767	90	13	p	p	NN
5767	90	14	)	)	-RRB-
5767	90	15	;	;	:
5767	90	16	and	and	CC
5767	90	17	c.	c.	VBP
5767	90	18	This	this	DT
5767	90	19	distribution	distribution	NN
5767	90	20	is	be	VBZ
5767	90	21	mathematically	mathematically	RB
5767	90	22	tractable	tractable	JJ
5767	90	23	.	.	.
5767	91	1	If	if	IN
5767	91	2	we	-PRON-	PRP
5767	91	3	proceed	proceed	VBP
5767	91	4	using	use	VBG
5767	91	5	the	the	DT
5767	91	6	above	above	JJ
5767	91	7	approximation	approximation	NN
5767	91	8	for	for	IN
5767	91	9	g(p	g(p	NNP
5767	91	10	)	)	-RRB-
5767	91	11	,	,	,
5767	91	12	we	-PRON-	PRP
5767	91	13	find	find	VBP
5767	91	14	:	:	:
5767	91	15	(	(	-LRB-
5767	91	16	i	i	NN
5767	91	17	)	)	-RRB-
5767	91	18	the	the	DT
5767	91	19	probability	probability	NN
5767	91	20	,	,	,
5767	91	21	P(n	P(n	NNP
5767	91	22	)	)	-RRB-
5767	91	23	,	,	,
5767	91	24	of	of	IN
5767	91	25	n	n	JJ
5767	91	26	items	item	NNS
5767	91	27	being	be	VBG
5767	91	28	retrieved	retrieve	VBN
5767	91	29	is	be	VBZ
5767	91	30	given	give	VBN
5767	91	31	by	by	IN
5767	91	32	1	1	CD
5767	91	33	.	.	.
5767	92	1	P(n	p(n	NN
5767	92	2	)	)	-RRB-
5767	92	3	=	=	NFP
5767	92	4	(	(	-LRB-
5767	92	5	-N	-N	NFP
5767	92	6	)	)	-RRB-
5767	92	7	~-+	~-+	NFP
5767	92	8	f3	f3	NN
5767	92	9	+	+	CC
5767	92	10	1~l(a	1~l(a	CD
5767	92	11	+	+	SYM
5767	92	12	n	n	NN
5767	92	13	)	)	-RRB-
5767	92	14	!	!	.
5767	93	1	(	(	-LRB-
5767	93	2	N-	n-	CD
5767	93	3	n	n	CC
5767	93	4	+	+	CC
5767	93	5	[	[	-LRB-
5767	93	6	3	3	CD
5767	93	7	)	)	-RRB-
5767	93	8	!	!	.
5767	94	1	n	n	NNP
5767	94	2	a!fJ	a!fJ	NNP
5767	94	3	!	!	.
5767	95	1	(	(	-LRB-
5767	95	2	a+fJ+N+l	a+fJ+N+l	NNP
5767	95	3	)	)	-RRB-
5767	95	4	!	!	.
5767	96	1	(	(	-LRB-
5767	96	2	ii	ii	NNP
5767	96	3	)	)	-RRB-
5767	96	4	the	the	DT
5767	96	5	expected	expected	JJ
5767	96	6	number	number	NN
5767	96	7	of	of	IN
5767	96	8	items	item	NNS
5767	96	9	retrieved	retrieve	VBN
5767	96	10	,	,	,
5767	96	11	E	e	NN
5767	96	12	,	,	,
5767	96	13	is	be	VBZ
5767	96	14	given	give	VBN
5767	96	15	by	by	IN
5767	96	16	a	a	DT
5767	96	17	+	+	SYM
5767	96	18	1	1	CD
5767	96	19	2	2	CD
5767	96	20	.	.	.
5767	97	1	E	E	NNP
5767	97	2	=	=	SYM
5767	97	3	=	=	NFP
5767	97	4	N	n	NN
5767	97	5	a	a	NN
5767	97	6	+	+	SYM
5767	97	7	{	{	-LRB-
5767	97	8	3	3	CD
5767	97	9	+	+	SYM
5767	97	10	2	2	CD
5767	97	11	;	;	:
5767	97	12	and	and	CC
5767	97	13	(	(	-LRB-
5767	97	14	iii	iii	NNP
5767	97	15	)	)	-RRB-
5767	97	16	the	the	DT
5767	97	17	variance	variance	NN
5767	97	18	,	,	,
5767	97	19	V	V	NNP
5767	97	20	,	,	,
5767	97	21	of	of	IN
5767	97	22	the	the	DT
5767	97	23	number	number	NN
5767	97	24	of	of	IN
5767	97	25	items	item	NNS
5767	97	26	retrieved	retrieve	VBN
5767	97	27	is	be	VBZ
5767	97	28	given	give	VBN
5767	97	29	by	by	IN
5767	97	30	_	_	NNP
5767	97	31	a+l	a+l	NNP
5767	97	32	{	{	-LRB-
5767	97	33	3	3	CD
5767	97	34	+	+	SYM
5767	97	35	1	1	CD
5767	97	36	N	n	CD
5767	97	37	3	3	CD
5767	97	38	·	·	NFP
5767	97	39	V	v	NN
5767	97	40	-	-	HYPH
5767	97	41	N	n	NN
5767	97	42	a	a	NN
5767	97	43	+	+	SYM
5767	97	44	f3	f3	NN
5767	97	45	+	+	CC
5767	97	46	2	2	CD
5767	97	47	a	a	NN
5767	97	48	+	+	SYM
5767	97	49	{	{	-LRB-
5767	97	50	3	3	CD
5767	97	51	+	+	SYM
5767	97	52	3	3	CD
5767	97	53	(	(	-LRB-
5767	97	54	1	1	CD
5767	97	55	+	+	SYM
5767	97	56	a.	a.	NN
5767	98	1	+	+	CC
5767	98	2	{	{	-LRB-
5767	98	3	3	3	CD
5767	98	4	+	+	SYM
5767	98	5	2	2	CD
5767	98	6	)	)	-RRB-
5767	98	7	·	·	NFP
5767	98	8	If	if	IN
5767	98	9	the	the	DT
5767	98	10	experiment	experiment	NN
5767	98	11	is	be	VBZ
5767	98	12	performed	perform	VBN
5767	98	13	on	on	IN
5767	98	14	a	a	DT
5767	98	15	small	small	JJ
5767	98	16	sample	sample	NN
5767	98	17	,	,	,
5767	98	18	the	the	DT
5767	98	19	expectation	expectation	NN
5767	98	20	and	and	CC
5767	98	21	variance	variance	NN
5767	98	22	can	can	MD
5767	98	23	be	be	VB
5767	98	24	computed	compute	VBN
5767	98	25	and	and	CC
5767	98	26	the	the	DT
5767	98	27	values	value	NNS
5767	98	28	of	of	IN
5767	98	29	a	a	DT
5767	98	30	and	and	CC
5767	98	31	f1	f1	NNP
5767	98	32	estimated	estimate	VBN
5767	98	33	from	from	IN
5767	98	34	the	the	DT
5767	98	35	relations	relation	NNS
5767	98	36	E	e	NN
5767	98	37	a	a	DT
5767	98	38	(	(	-LRB-
5767	98	39	1	1	CD
5767	98	40	-	-	HYPH
5767	98	41	-	-	HYPH
5767	98	42	)	)	-RRB-
5767	98	43	+	+	CC
5767	98	44	1	1	CD
5767	98	45	4	4	CD
5767	98	46	.	.	.
5767	98	47	f1==	f1==	NFP
5767	98	48	N	n	NN
5767	98	49	2	2	CD
5767	98	50	,	,	,
5767	98	51	and	and	CC
5767	98	52	E	e	NN
5767	98	53	N	n	NN
5767	98	54	v	v	NN
5767	98	55	E-	e-	NN
5767	98	56	N	n	NN
5767	98	57	E	e	NN
5767	98	58	1	1	CD
5767	98	59	-N	-n	SYM
5767	98	60	5	5	CD
5767	98	61	.	.	.
5767	98	62	a.	a.	NN
5767	98	63	v	v	NNP
5767	98	64	-	-	HYPH
5767	98	65	1	1	CD
5767	98	66	E	e	NN
5767	98	67	l	l	NN
5767	98	68	E	e	NN
5767	98	69	1	1	CD
5767	98	70	-	-	HYPH
5767	98	71	-	-	:
5767	98	72	N	n	NN
5767	98	73	Usually	usually	RB
5767	98	74	~	~	NFP
5767	98	75	will	will	MD
5767	98	76	be	be	VB
5767	98	77	much	much	RB
5767	98	78	smaller	small	JJR
5767	98	79	than	than	IN
5767	98	80	one	one	CD
5767	98	81	;	;	:
5767	98	82	in	in	IN
5767	98	83	this	this	DT
5767	98	84	case	case	NN
5767	98	85	we	-PRON-	PRP
5767	98	86	may	may	MD
5767	98	87	use	use	VB
5767	98	88	the	the	DT
5767	98	89	approximations	approximation	NNS
5767	98	90	:	:	:
5767	98	91	N	n	NN
5767	98	92	4	4	CD
5767	98	93	'	'	''
5767	98	94	.	.	.
5767	99	1	f3	f3	NNP
5767	99	2	=	=	NFP
5767	99	3	(	(	-LRB-
5767	99	4	a+	a+	NNP
5767	99	5	I)E	I)E	NNP
5767	99	6	,	,	,
5767	99	7	and	and	CC
5767	99	8	E	e	NN
5767	99	9	5	5	CD
5767	99	10	'	'	''
5767	99	11	.	.	.
5767	100	1	E	E	NNP
5767	100	2	1	1	CD
5767	100	3	a=	a=	NNP
5767	100	4	N-	n-	CD
5767	100	5	.	.	.
5767	101	1	Once	once	RB
5767	101	2	a	a	DT
5767	101	3	and	and	CC
5767	101	4	f1	f1	NNP
5767	101	5	have	have	VBP
5767	101	6	been	be	VBN
5767	101	7	evaluated	evaluate	VBN
5767	101	8	,	,	,
5767	101	9	we	-PRON-	PRP
5767	101	10	can	can	MD
5767	101	11	compute	compute	VB
5767	101	12	the	the	DT
5767	101	13	probabilities	probability	NNS
5767	101	14	P	P	NNP
5767	101	15	(	(	-LRB-
5767	101	16	n	n	NN
5767	101	17	)	)	-RRB-
5767	101	18	for	for	IN
5767	101	19	files	file	NNS
5767	101	20	of	of	IN
5767	101	21	arbitrary	arbitrary	JJ
5767	101	22	size	size	NN
5767	101	23	,	,	,
5767	101	24	and	and	CC
5767	101	25	with	with	IN
5767	101	26	these	these	DT
5767	101	27	values	value	NNS
5767	101	28	we	-PRON-	PRP
5767	101	29	can	can	MD
5767	101	30	make	make	VB
5767	101	31	as-	as-	JJ
5767	101	32	sertions	sertion	NNS
5767	101	33	regarding	regard	VBG
5767	101	34	the	the	DT
5767	101	35	probability	probability	NN
5767	101	36	of	of	IN
5767	101	37	,	,	,
5767	101	38	say	say	VB
5767	101	39	,	,	,
5767	101	40	more	more	JJR
5767	101	41	than	than	IN
5767	101	42	30	30	CD
5767	101	43	items	item	NNS
5767	101	44	being	be	VBG
5767	101	45	re-	re-	RB
5767	101	46	trieved	trieve	VBN
5767	101	47	.	.	.
5767	102	1	A	a	DT
5767	102	2	relation	relation	NN
5767	102	3	that	that	WDT
5767	102	4	can	can	MD
5767	102	5	be	be	VB
5767	102	6	derived	derive	VBN
5767	102	7	from	from	IN
5767	102	8	Formula	Formula	NNP
5767	102	9	1	1	CD
5767	102	10	and	and	CC
5767	102	11	may	may	MD
5767	102	12	be	be	VB
5767	102	13	of	of	IN
5767	102	14	use	use	NN
5767	102	15	when	when	WRB
5767	102	16	comparing	compare	VBG
5767	102	17	this	this	DT
5767	102	18	model	model	NN
5767	102	19	with	with	IN
5767	102	20	experiment	experiment	NN
5767	102	21	is	be	VBZ
5767	102	22	:	:	:
5767	102	23	P(n	p(n	NN
5767	102	24	)	)	-RRB-
5767	102	25	I	i	NN
5767	102	26	+	+	CC
5767	102	27	{	{	-LRB-
5767	102	28	3	3	CD
5767	102	29	N	n	CD
5767	102	30	-	-	HYPH
5767	102	31	n	n	NN
5767	102	32	=	=	SYM
5767	102	33	1	1	CD
5767	102	34	+	+	CC
5767	102	35	a	a	DT
5767	102	36	n	n	NN
5767	102	37	+	+	CC
5767	102	38	1	1	CD
5767	102	39	P(n	p(n	CD
5767	102	40	+	+	SYM
5767	102	41	1	1	CD
5767	102	42	)	)	-RRB-
5767	102	43	Statistical	statistical	JJ
5767	102	44	Behavior	Behavior	NNP
5767	102	45	of	of	IN
5767	102	46	Search	Search	NNP
5767	102	47	Keys/	Keys/	NNP
5767	102	48	BOOKSTEIN	BOOKSTEIN	NNP
5767	102	49	115	115	CD
5767	102	50	The	the	DT
5767	102	51	probability	probability	NN
5767	102	52	of	of	IN
5767	102	53	zero	zero	CD
5767	102	54	retrievals	retrieval	NNS
5767	102	55	is	be	VBZ
5767	102	56	likely	likely	JJ
5767	102	57	to	to	TO
5767	102	58	be	be	VB
5767	102	59	an	an	DT
5767	102	60	extraordinary	extraordinary	JJ
5767	102	61	point	point	NN
5767	102	62	in	in	IN
5767	102	63	the	the	DT
5767	102	64	distributions	distribution	NNS
5767	102	65	g	g	NN
5767	102	66	(	(	-LRB-
5767	102	67	p	p	NN
5767	102	68	)	)	-RRB-
5767	102	69	and	and	CC
5767	102	70	P	p	NN
5767	102	71	(	(	-LRB-
5767	102	72	n	n	NNP
5767	102	73	)	)	-RRB-
5767	102	74	since	since	IN
5767	102	75	it	-PRON-	PRP
5767	102	76	is	be	VBZ
5767	102	77	influenced	influence	VBN
5767	102	78	by	by	IN
5767	102	79	the	the	DT
5767	102	80	knowledge	knowledge	NN
5767	102	81	that	that	IN
5767	102	82	a	a	DT
5767	102	83	user	user	NN
5767	102	84	may	may	MD
5767	102	85	have	have	VB
5767	102	86	of	of	IN
5767	102	87	the	the	DT
5767	102	88	collection	collection	NN
5767	102	89	;	;	:
5767	102	90	this	this	DT
5767	102	91	effect	effect	NN
5767	102	92	is	be	VBZ
5767	102	93	likely	likely	JJ
5767	102	94	to	to	TO
5767	102	95	be	be	VB
5767	102	96	encountered	encounter	VBN
5767	102	97	in	in	IN
5767	102	98	a	a	DT
5767	102	99	sampling	sampling	NN
5767	102	100	process	process	NN
5767	102	101	in	in	IN
5767	102	102	which	which	WDT
5767	102	103	the	the	DT
5767	102	104	requests	request	NNS
5767	102	105	have	have	VBP
5767	102	106	to	to	TO
5767	102	107	be	be	VB
5767	102	108	generated	generate	VBN
5767	102	109	artificially	artificially	RB
5767	102	110	.	.	.
5767	103	1	In	in	IN
5767	103	2	such	such	JJ
5767	103	3	cases	case	NNS
5767	103	4	it	-PRON-	PRP
5767	103	5	would	would	MD
5767	103	6	be	be	VB
5767	103	7	advisable	advisable	JJ
5767	103	8	to	to	TO
5767	103	9	treat	treat	VB
5767	103	10	P	p	NN
5767	103	11	(	(	-LRB-
5767	103	12	0	0	NFP
5767	103	13	)	)	-RRB-
5767	103	14	as	as	IN
5767	103	15	an	an	DT
5767	103	16	empirically	empirically	RB
5767	103	17	derived	derive	VBN
5767	103	18	parameter	parameter	NN
5767	103	19	,	,	,
5767	103	20	(	(	-LRB-
5767	103	21	)	)	-RRB-
5767	103	22	,	,	,
5767	103	23	and	and	CC
5767	103	24	use	use	VB
5767	103	25	the	the	DT
5767	103	26	modified	modify	VBN
5767	103	27	formula	formula	NN
5767	103	28	{	{	-LRB-
5767	103	29	(	(	-LRB-
5767	103	30	Jifn	Jifn	NNP
5767	103	31	=	=	SYM
5767	103	32	O	o	NN
5767	103	33	6	6	CD
5767	103	34	.	.	.
5767	104	1	P	p	NN
5767	104	2	'	'	''
5767	104	3	(	(	-LRB-
5767	104	4	n	n	NN
5767	104	5	)	)	-RRB-
5767	104	6	=	=	NFP
5767	104	7	(	(	-LRB-
5767	104	8	1	1	CD
5767	104	9	-	-	HYPH
5767	104	10	fJ	fj	NN
5767	104	11	)	)	-RRB-
5767	104	12	1	1	CD
5767	104	13	~(;~O	~(;~o	NN
5767	104	14	)	)	-RRB-
5767	104	15	if	if	IN
5767	104	16	n	n	NN
5767	104	17	:	:	:
5767	104	18	:	:	:
5767	104	19	1=	1=	CD
5767	104	20	0	0	CD
5767	104	21	.	.	.
5767	105	1	The	the	DT
5767	105	2	value	value	NN
5767	105	3	of	of	IN
5767	105	4	(	(	-LRB-
5767	105	5	)	)	-RRB-
5767	105	6	can	can	MD
5767	105	7	be	be	VB
5767	105	8	estimated	estimate	VBN
5767	105	9	by	by	IN
5767	105	10	the	the	DT
5767	105	11	fraction	fraction	NN
5767	105	12	of	of	IN
5767	105	13	requests	request	NNS
5767	105	14	retrieving	retrieve	VBG
5767	105	15	zero	zero	CD
5767	105	16	items	item	NNS
5767	105	17	;	;	:
5767	105	18	for	for	IN
5767	105	19	sampling	sample	VBG
5767	105	20	techniques	technique	NNS
5767	105	21	using	use	VBG
5767	105	22	only	only	JJ
5767	105	23	productive	productive	JJ
5767	105	24	requests	request	NNS
5767	105	25	,	,	,
5767	105	26	(	(	-LRB-
5767	105	27	)	)	-RRB-
5767	105	28	will	will	MD
5767	105	29	be	be	VB
5767	105	30	zero	zero	CD
5767	105	31	.	.	.
5767	106	1	a.	a.	NNP
5767	106	2	and	and	CC
5767	106	3	f3	f3	NNP
5767	106	4	can	can	MD
5767	106	5	be	be	VB
5767	106	6	calculated	calculate	VBN
5767	106	7	as	as	IN
5767	106	8	before	before	RB
5767	106	9	from	from	IN
5767	106	10	the	the	DT
5767	106	11	mean	mean	NN
5767	106	12	and	and	CC
5767	106	13	variance	variance	NN
5767	106	14	of	of	IN
5767	106	15	the	the	DT
5767	106	16	sample	sample	NN
5767	106	17	.	.	.
5767	107	1	CONCLUSION	conclusion	NN
5767	107	2	The	the	DT
5767	107	3	above	above	JJ
5767	107	4	discussion	discussion	NN
5767	107	5	is	be	VBZ
5767	107	6	intended	intend	VBN
5767	107	7	as	as	IN
5767	107	8	an	an	DT
5767	107	9	attempt	attempt	NN
5767	107	10	to	to	TO
5767	107	11	provide	provide	VB
5767	107	12	some	some	DT
5767	107	13	theoreti-	theoreti-	JJ
5767	107	14	cal	cal	NN
5767	107	15	understanding	understanding	NN
5767	107	16	of	of	IN
5767	107	17	the	the	DT
5767	107	18	puzzling	puzzle	VBG
5767	107	19	behavior	behavior	NN
5767	107	20	discovered	discover	VBN
5767	107	21	in	in	IN
5767	107	22	the	the	DT
5767	107	23	use	use	NN
5767	107	24	of	of	IN
5767	107	25	search	search	NN
5767	107	26	keys	key	NNS
5767	107	27	and	and	CC
5767	107	28	also	also	RB
5767	107	29	to	to	TO
5767	107	30	provide	provide	VB
5767	107	31	some	some	DT
5767	107	32	guide	guide	NN
5767	107	33	for	for	IN
5767	107	34	those	those	DT
5767	107	35	experimenting	experiment	VBG
5767	107	36	with	with	IN
5767	107	37	samples	sample	NNS
5767	107	38	of	of	IN
5767	107	39	such	such	JJ
5767	107	40	files	file	NNS
5767	107	41	.	.	.
5767	108	1	We	-PRON-	PRP
5767	108	2	do	do	VBP
5767	108	3	,	,	,
5767	108	4	however	however	RB
5767	108	5	,	,	,
5767	108	6	urge	urge	NN
5767	108	7	caution	caution	NN
5767	108	8	for	for	IN
5767	108	9	the	the	DT
5767	108	10	latter	latter	JJ
5767	108	11	uses	use	NNS
5767	108	12	.	.	.
5767	109	1	An	an	DT
5767	109	2	analysis	analysis	NN
5767	109	3	similar	similar	JJ
5767	109	4	to	to	IN
5767	109	5	the	the	DT
5767	109	6	above	above	RB
5767	109	7	can	can	MD
5767	109	8	be	be	VB
5767	109	9	useful	useful	JJ
5767	109	10	under	under	IN
5767	109	11	several	several	JJ
5767	109	12	different	different	JJ
5767	109	13	circumstances	circumstance	NNS
5767	109	14	,	,	,
5767	109	15	such	such	JJ
5767	109	16	as	as	IN
5767	109	17	:	:	:
5767	109	18	determining	determine	VBG
5767	109	19	the	the	DT
5767	109	20	future	future	JJ
5767	109	21	behavior	behavior	NN
5767	109	22	expected	expect	VBN
5767	109	23	of	of	IN
5767	109	24	a	a	DT
5767	109	25	search	search	NN
5767	109	26	key	key	NN
5767	109	27	in	in	IN
5767	109	28	a	a	DT
5767	109	29	single	single	JJ
5767	109	30	library	library	NN
5767	109	31	as	as	IN
5767	109	32	the	the	DT
5767	109	33	collection	collection	NN
5767	109	34	grows	grow	VBZ
5767	109	35	;	;	:
5767	109	36	determining	determine	VBG
5767	109	37	the	the	DT
5767	109	38	be-	be-	JJ
5767	109	39	havior	havior	NNP
5767	109	40	for	for	IN
5767	109	41	one	one	CD
5767	109	42	library	library	NN
5767	109	43	based	base	VBN
5767	109	44	upon	upon	IN
5767	109	45	experiments	experiment	NNS
5767	109	46	conducted	conduct	VBN
5767	109	47	on	on	IN
5767	109	48	a	a	DT
5767	109	49	different	different	JJ
5767	109	50	but	but	CC
5767	109	51	similar	similar	JJ
5767	109	52	library	library	NN
5767	109	53	;	;	:
5767	109	54	and	and	CC
5767	109	55	extrapolating	extrapolate	VBG
5767	109	56	from	from	IN
5767	109	57	the	the	DT
5767	109	58	performance	performance	NN
5767	109	59	of	of	IN
5767	109	60	a	a	DT
5767	109	61	search	search	NN
5767	109	62	key	key	NN
5767	109	63	in	in	IN
5767	109	64	a	a	DT
5767	109	65	sample	sample	NN
5767	109	66	of	of	IN
5767	109	67	the	the	DT
5767	109	68	collection	collection	NN
5767	109	69	to	to	IN
5767	109	70	its	-PRON-	PRP$
5767	109	71	pedormance	pedormance	NN
5767	109	72	in	in	IN
5767	109	73	the	the	DT
5767	109	74	full	full	JJ
5767	109	75	collection	collection	NN
5767	109	76	.	.	.
5767	110	1	If	if	IN
5767	110	2	one	one	CD
5767	110	3	wishes	wish	VBZ
5767	110	4	to	to	TO
5767	110	5	compare	compare	VB
5767	110	6	two	two	CD
5767	110	7	different	different	JJ
5767	110	8	libraries	library	NNS
5767	110	9	,	,	,
5767	110	10	one	one	PRP
5767	110	11	can	can	MD
5767	110	12	note	note	VB
5767	110	13	that	that	IN
5767	110	14	as	as	RB
5767	110	15	far	far	RB
5767	110	16	as	as	IN
5767	110	17	search	search	NN
5767	110	18	key	key	JJ
5767	110	19	values	value	NNS
5767	110	20	are	be	VBP
5767	110	21	concerned	concern	VBN
5767	110	22	,	,	,
5767	110	23	a	a	DT
5767	110	24	particular	particular	JJ
5767	110	25	library	library	NN
5767	110	26	's	's	POS
5767	110	27	collection	collection	NN
5767	110	28	can	can	MD
5767	110	29	be	be	VB
5767	110	30	thought	think	VBN
5767	110	31	of	of	IN
5767	110	32	as	as	IN
5767	110	33	a	a	DT
5767	110	34	random	random	JJ
5767	110	35	sample	sample	NN
5767	110	36	of	of	IN
5767	110	37	the	the	DT
5767	110	38	larger	large	JJR
5767	110	39	population	population	NN
5767	110	40	from	from	IN
5767	110	41	which	which	WDT
5767	110	42	it	-PRON-	PRP
5767	110	43	selects	select	VBZ
5767	110	44	its	-PRON-	PRP$
5767	110	45	material	material	NN
5767	110	46	,	,	,
5767	110	47	and	and	CC
5767	110	48	accordingly	accordingly	RB
5767	110	49	the	the	DT
5767	110	50	formula	formula	NN
5767	110	51	for	for	IN
5767	110	52	P	p	NN
5767	110	53	(	(	-LRB-
5767	110	54	n	n	NN
5767	110	55	)	)	-RRB-
5767	110	56	should	should	MD
5767	110	57	be	be	VB
5767	110	58	valid	valid	JJ
5767	110	59	.	.	.
5767	111	1	In	in	IN
5767	111	2	this	this	DT
5767	111	3	case	case	NN
5767	111	4	,	,	,
5767	111	5	if	if	IN
5767	111	6	two	two	CD
5767	111	7	different	different	JJ
5767	111	8	collections	collection	NNS
5767	111	9	are	be	VBP
5767	111	10	drawn	draw	VBN
5767	111	11	from	from	IN
5767	111	12	the	the	DT
5767	111	13	same	same	JJ
5767	111	14	population	population	NN
5767	111	15	,	,	,
5767	111	16	the	the	DT
5767	111	17	g	g	NN
5767	111	18	(	(	-LRB-
5767	111	19	p	p	NN
5767	111	20	)	)	-RRB-
5767	111	21	refers	refer	VBZ
5767	111	22	to	to	IN
5767	111	23	this	this	DT
5767	111	24	population	population	NN
5767	111	25	and	and	CC
5767	111	26	the	the	DT
5767	111	27	libraries	library	NNS
5767	111	28	are	be	VBP
5767	111	29	distinguished	distinguish	VBN
5767	111	30	by	by	IN
5767	111	31	the	the	DT
5767	111	32	parameter	parameter	NN
5767	111	33	N	n	NN
5767	111	34	;	;	:
5767	111	35	when	when	WRB
5767	111	36	we	-PRON-	PRP
5767	111	37	are	be	VBP
5767	111	38	considering	consider	VBG
5767	111	39	samples	sample	NNS
5767	111	40	from	from	IN
5767	111	41	a	a	DT
5767	111	42	single	single	JJ
5767	111	43	library	library	NN
5767	111	44	,	,	,
5767	111	45	then	then	RB
5767	111	46	N	n	NN
5767	111	47	is	be	VBZ
5767	111	48	the	the	DT
5767	111	49	sample	sample	NN
5767	111	50	size	size	NN
5767	111	51	and	and	CC
5767	111	52	g	g	NNP
5767	111	53	(	(	-LRB-
5767	111	54	p	p	NN
5767	111	55	)	)	-RRB-
5767	111	56	refers	refer	VBZ
5767	111	57	to	to	IN
5767	111	58	the	the	DT
5767	111	59	library	library	NN
5767	111	60	itself	-PRON-	PRP
5767	111	61	.	.	.
5767	112	1	No	no	DT
5767	112	2	theoretical	theoretical	JJ
5767	112	3	basis	basis	NN
5767	112	4	exists	exist	VBZ
5767	112	5	at	at	IN
5767	112	6	present	present	NN
5767	112	7	for	for	IN
5767	112	8	estimating	estimate	VBG
5767	112	9	to	to	IN
5767	112	10	what	what	WDT
5767	112	11	extent	extent	NN
5767	112	12	the	the	DT
5767	112	13	populations	population	NNS
5767	112	14	being	be	VBG
5767	112	15	considered	consider	VBN
5767	112	16	depend	depend	VBP
5767	112	17	upon	upon	IN
5767	112	18	the	the	DT
5767	112	19	type	type	NN
5767	112	20	of	of	IN
5767	112	21	library	library	NN
5767	112	22	,	,	,
5767	112	23	if	if	IN
5767	112	24	any	any	DT
5767	112	25	,	,	,
5767	112	26	so	so	RB
5767	112	27	this	this	DT
5767	112	28	problem	problem	NN
5767	112	29	must	must	MD
5767	112	30	be	be	VB
5767	112	31	dealt	deal	VBN
5767	112	32	with	with	IN
5767	112	33	empirically	empirically	RB
5767	112	34	.	.	.
5767	113	1	We	-PRON-	PRP
5767	113	2	have	have	VBP
5767	113	3	assumed	assume	VBN
5767	113	4	here	here	RB
5767	113	5	that	that	IN
5767	113	6	these	these	DT
5767	113	7	populations	population	NNS
5767	113	8	are	be	VBP
5767	113	9	similar	similar	JJ
5767	113	10	with	with	IN
5767	113	11	regard	regard	NN
5767	113	12	to	to	TO
5767	113	13	search	search	VB
5767	113	14	key	key	JJ
5767	113	15	values	value	NNS
5767	113	16	.	.	.
5767	114	1	Should	Should	MD
5767	114	2	these	these	DT
5767	114	3	populations	population	NNS
5767	114	4	in	in	IN
5767	114	5	fact	fact	NN
5767	114	6	vary	vary	VBP
5767	114	7	,	,	,
5767	114	8	it	-PRON-	PRP
5767	114	9	is	be	VBZ
5767	114	10	possible	possible	JJ
5767	114	11	that	that	IN
5767	114	12	they	-PRON-	PRP
5767	114	13	can	can	MD
5767	114	14	be	be	VB
5767	114	15	broken	break	VBN
5767	114	16	down	down	RP
5767	114	17	,	,	,
5767	114	18	e.g.	e.g.	RB
5767	114	19	,	,	,
5767	114	20	by	by	IN
5767	114	21	language	language	NN
5767	114	22	,	,	,
5767	114	23	into	into	IN
5767	114	24	subpopulations	subpopulation	NNS
5767	114	25	that	that	WDT
5767	114	26	are	be	VBP
5767	114	27	stable	stable	JJ
5767	114	28	and	and	CC
5767	114	29	for	for	IN
5767	114	30	each	each	DT
5767	114	31	of	of	IN
5767	114	32	which	which	WDT
5767	114	33	the	the	DT
5767	114	34	analysis	analysis	NN
5767	114	35	is	be	VBZ
5767	114	36	valid	valid	JJ
5767	114	37	.	.	.
5767	115	1	ACKNOWLEDGMENTS	acknowledgment	NNS
5767	115	2	This	this	DT
5767	115	3	work	work	NN
5767	115	4	was	be	VBD
5767	115	5	made	make	VBN
5767	115	6	possible	possible	JJ
5767	115	7	by	by	IN
5767	115	8	CLR/	CLR/	NNP
5767	115	9	NEH	NEH	NNP
5767	115	10	Grant	Grant	NNP
5767	115	11	No	no	NN
5767	115	12	.	.	.
5767	116	1	E0	E0	NNP
5767	116	2	-	-	HYPH
5767	116	3	262	262	CD
5767	116	4	-	-	HYPH
5767	116	5	70	70	CD
5767	116	6	-	-	HYPH
5767	116	7	4658	4658	CD
5767	116	8	.	.	.
5767	117	1	I	-PRON-	PRP
5767	117	2	would	would	MD
5767	117	3	like	like	VB
5767	117	4	to	to	TO
5767	117	5	express	express	VB
5767	117	6	my	-PRON-	PRP$
5767	117	7	gratitude	gratitude	NN
5767	117	8	to	to	IN
5767	117	9	members	member	NNS
5767	117	10	of	of	IN
5767	117	11	the	the	DT
5767	117	12	University	University	NNP
5767	117	13	of	of	IN
5767	117	14	Chicago	Chicago	NNP
5767	117	15	Systems	Systems	NNPS
5767	117	16	Develop-	Develop-	NNP
5767	117	17	ment	ment	JJ
5767	117	18	Office	Office	NNP
5767	117	19	for	for	IN
5767	117	20	their	-PRON-	PRP$
5767	117	21	many	many	JJ
5767	117	22	comments	comment	NNS
5767	117	23	and	and	CC
5767	117	24	suggestions	suggestion	NNS
5767	117	25	on	on	IN
5767	117	26	this	this	DT
5767	117	27	work	work	NN
5767	117	28	.	.	.
5767	118	1	I	-PRON-	PRP
5767	118	2	;	;	:
5767	118	3	I	-PRON-	PRP
5767	118	4	ll6	ll6	VBP
5767	118	5	Journal	Journal	NNP
5767	118	6	of	of	IN
5767	118	7	Library	Library	NNP
5767	118	8	Automation	Automation	NNP
5767	118	9	Vol	Vol	NNP
5767	118	10	.	.	.
5767	119	1	6/	6/	CD
5767	119	2	2	2	CD
5767	119	3	June	June	NNP
5767	119	4	1973	1973	CD
5767	119	5	REFERENCES	reference	NNS
5767	119	6	I.	I.	NNP
5767	119	7	Frederick	Frederick	NNP
5767	119	8	G.	G.	NNP
5767	119	9	Kilgour	Kilgour	NNP
5767	119	10	,	,	,
5767	119	11	Philip	Philip	NNP
5767	119	12	L.	L.	NNP
5767	119	13	Long	Long	NNP
5767	119	14	,	,	,
5767	119	15	Eugene	Eugene	NNP
5767	119	16	B.	B.	NNP
5767	119	17	Leiderman	Leiderman	NNP
5767	119	18	,	,	,
5767	119	19	and	and	CC
5767	119	20	Alan	Alan	NNP
5767	119	21	L.	L.	NNP
5767	119	22	Landgraf	Landgraf	NNP
5767	119	23	,	,	,
5767	119	24	"	"	``
5767	119	25	Title	title	NN
5767	119	26	-	-	HYPH
5767	119	27	Only	Only	NNP
5767	119	28	Entries	Entries	NNPS
5767	119	29	Retrieved	retrieve	VBN
5767	119	30	by	by	IN
5767	119	31	Use	Use	NNP
5767	119	32	of	of	IN
5767	119	33	Truncated	Truncated	NNP
5767	119	34	Search	Search	NNP
5767	119	35	Key	Key	NNP
5767	119	36	,	,	,
5767	119	37	"	"	''
5767	119	38	Journal	Journal	NNP
5767	119	39	of	of	IN
5767	119	40	Library	Library	NNP
5767	119	41	Automation	Automation	NNP
5767	119	42	4:207	4:207	CD
5767	119	43	-	-	SYM
5767	119	44	10	10	CD
5767	119	45	(	(	-LRB-
5767	119	46	Dec.	December	NNP
5767	119	47	1971	1971	CD
5767	119	48	)	)	-RRB-
5767	119	49	.	.	.
5767	120	1	2	2	LS
5767	120	2	.	.	.
5767	121	1	A.	A.	NNP
5767	121	2	Bookstein	Bookstein	NNP
5767	121	3	,	,	,
5767	121	4	"	"	``
5767	121	5	Double	double	JJ
5767	121	6	Hashing	hashing	NN
5767	121	7	,	,	,
5767	121	8	"	"	''
5767	121	9	Journal	Journal	NNP
5767	121	10	of	of	IN
5767	121	11	the	the	DT
5767	121	12	American	American	NNP
5767	121	13	Society	Society	NNP
5767	121	14	for	for	IN
5767	121	15	Information	Information	NNP
5767	121	16	Science	Science	NNP
5767	121	17	23:402	23:402	CD
5767	121	18	-	-	SYM
5767	121	19	25	25	CD
5767	121	20	(	(	-LRB-
5767	121	21	Nov.-Dec	nov.-dec	NN
5767	121	22	.	.	.
5767	121	23	1972	1972	CD
5767	121	24	)	)	-RRB-
5767	121	25	.	.	.
5767	122	1	3	3	LS
5767	122	2	.	.	.
5767	123	1	A.	A.	NNP
5767	123	2	Bookstein	Bookstein	NNP
5767	123	3	,	,	,
5767	123	4	"	"	``
5767	123	5	Hash	Hash	NNP
5767	123	6	Coding	Coding	NNP
5767	123	7	with	with	IN
5767	123	8	a	a	DT
5767	123	9	Non	non	JJ
5767	123	10	-	-	JJ
5767	123	11	Unique	unique	JJ
5767	123	12	Search	Search	NNP
5767	123	13	Key	Key	NNP
5767	123	14	,	,	,
5767	123	15	"	"	''
5767	123	16	to	to	TO
5767	123	17	be	be	VB
5767	123	18	published	publish	VBN
5767	123	19	in	in	IN
5767	123	20	the	the	DT
5767	123	21	Journal	Journal	NNP
5767	123	22	of	of	IN
5767	123	23	American	American	NNP
5767	123	24	Society	Society	NNP
5767	123	25	for	for	IN
5767	123	26	Information	Information	NNP
5767	123	27	Science	Science	NNP
5767	123	28	.	.	.
5767	124	1	4	4	LS
5767	124	2	.	.	.
5767	125	1	Frederick	Frederick	NNP
5767	125	2	G.	G.	NNP
5767	125	3	Kilgour	Kilgour	NNP
5767	125	4	,	,	,
5767	125	5	Philip	Philip	NNP
5767	125	6	L.	L.	NNP
5767	125	7	Long	Long	NNP
5767	125	8	,	,	,
5767	125	9	Eugene	Eugene	NNP
5767	125	10	B.	B.	NNP
5767	125	11	Leiderman	Leiderman	NNP
5767	125	12	,	,	,
5767	125	13	and	and	CC
5767	125	14	AJan	AJan	NNP
5767	125	15	L.	L.	NNP
5767	125	16	Landgraf	Landgraf	NNP
5767	125	17	,	,	,
5767	125	18	"	"	``
5767	125	19	Retrieval	retrieval	NN
5767	125	20	of	of	IN
5767	125	21	Bibliographic	Bibliographic	NNP
5767	125	22	Entries	Entries	NNP
5767	125	23	from	from	IN
5767	125	24	a	a	DT
5767	125	25	Name	Name	NNP
5767	125	26	-	-	HYPH
5767	125	27	Title	Title	NNP
5767	125	28	Catalog	Catalog	NNP
5767	125	29	by	by	IN
5767	125	30	Use	Use	NNP
5767	125	31	of	of	IN
5767	125	32	Truncated	Truncated	NNP
5767	125	33	Search	Search	NNP
5767	125	34	Keys	Keys	NNPS
5767	125	35	.	.	.
5767	125	36	"	"	''
5767	126	1	preprint	preprint	NN
5767	126	2	.	.	.
5767	127	1	5	5	CD
5767	127	2	.	.	.
5767	128	1	Kilgour	Kilgour	NNP
5767	128	2	,	,	,
5767	128	3	Long	Long	NNP
5767	128	4	,	,	,
5767	128	5	Leiderman	Leiderman	NNP
5767	128	6	,	,	,
5767	128	7	and	and	CC
5767	128	8	Landgraf	Landgraf	NNP
5767	128	9	,	,	,
5767	128	10	"	"	``
5767	128	11	Title	title	NN
5767	128	12	-	-	HYPH
5767	128	13	Only	Only	NNP
5767	128	14	Entries	Entries	NNPS
5767	128	15	,	,	,
5767	128	16	"	"	''
5767	128	17	p.209	p.209	NNP
5767	128	18	-	-	HYPH
5767	128	19	10	10	CD
5767	128	20	.	.	.
5767	129	1	6	6	CD
5767	129	2	.	.	.
5767	130	1	Gerry	Gerry	NNP
5767	130	2	P.	P.	NNP
5767	130	3	Guthrie	Guthrie	NNP
5767	130	4	and	and	CC
5767	130	5	Steven	Steven	NNP
5767	130	6	D.	D.	NNP
5767	130	7	Slifko	Slifko	NNP
5767	130	8	,	,	,
5767	130	9	"	"	``
5767	130	10	Analysis	Analysis	NNP
5767	130	11	of	of	IN
5767	130	12	Search	Search	NNP
5767	130	13	Key	Key	NNP
5767	130	14	Retrieval	Retrieval	NNP
5767	130	15	on	on	IN
5767	130	16	a	a	DT
5767	130	17	Large	large	JJ
5767	130	18	Bibliographic	Bibliographic	NNP
5767	130	19	File	File	NNP
5767	130	20	,	,	,
5767	130	21	"	"	''
5767	130	22	Journal	Journal	NNP
5767	130	23	of	of	IN
5767	130	24	Library	Library	NNP
5767	130	25	Automation	Automation	NNP
5767	130	26	5:96	5:96	NNP
5767	130	27	-	-	SYM
5767	130	28	100	100	CD
5767	130	29	(	(	-LRB-
5767	130	30	June	June	NNP
5767	130	31	1972	1972	CD
5767	130	32	)	)	-RRB-
5767	130	33	.	.	.