id	author	title	date	pages	extension	mime	words	sentences	flesch	summary	cache	txt
work_3iywiieamja5tayrhj5mjsm6gy	Hongsong Li	Data-Driven Metaphor Recognition and Explanation	2013	12	.pdf	application/pdf	7312	692	71	To our knowledge, this is the first purely datadriven approach of probabilistic metaphor acquisition, recognition, and explanation. Existing work on metaphor recognition and interpretation can be divided into two categories: contextoriented and knowledge-driven. the selection association) is widely used in more recent approaches for metaphor recognition and interpretation (Mason, 2004; Shutova, 2010; Shutova et For example, Mason (2004) learns domain-specific selectional preferences and use them to find mappings between concepts from different domains. metaphors by focusing on the nearby verbs of a potential source or target concept. from human-curated knowledge bases like WordNet, known metaphor and idiom sets. more candidate metaphor pairs from billions of sentences in the web corpus: That is, pairs extracted by the "is a" pattern contains at least two types of relations: the literal isA relations and the metaphor relations. We label a sentence in the set as a metaphor if the two nouns connected by BE do not actually have isA relation; or if	./cache/work_3iywiieamja5tayrhj5mjsm6gy.pdf	./txt/work_3iywiieamja5tayrhj5mjsm6gy.txt