Expressing Implicit Semantic Relations without Supervision

Download
  1. (PDF, 247 KB)
AuthorSearch for:
TypeArticle
ConferenceProceedings of the 21rst International Committee on Computational Linguistics and the 44th Annual Meeting of the Association for Computational Linguistics (COLING/ACL 2006), July 17-21, 2006., Sydney, Australia
AbstractWe present an unsupervised learning algorithm that mines large text corpora for patterns that express implicit semantic relations. For a given input word pair <em>X:Y</em> with some unspecified semantic relations, the corresponding output list of patterns (P<sub><em>1</em></sub>,…,P<sub><em>m</em></sub>)is ranked according to how well each pattern <em>P<sub>i</sub></em> expresses the relations between<em> X </em>and <em>Y</em> . For example, given ostrich = <em>X</em> and bird = <em>Y</em> , the two highest ranking output patterns are "<em>X</em> is the largest <em>Y</em>" and "<em>Y</em> such as the <em>X</em>". The output patterns are intended to be useful for finding further pairs with the same relations, to support the construction of lexicons, ontologies, and semantic networks. The patterns are sorted by <em>pertinence</em>, where the pertinence of a pattern <em>P<sub>i</sub></em> for a word pair <em>X:Y</em> is the expected relational similarity between the given pair and typical pairs for <em>P<sub>i</sub></em> . The algorithm is empirically evaluated on two tasks, solving multiple-choice SAT word analogy questions and classifying semantic relations in noun-modifier pairs. On both tasks, the algorithm achieves state-of- the-art results, performing significantly better than several alternative pattern ranking algorithms, based on tf-idf.
Publication date
LanguageEnglish
AffiliationNRC Institute for Information Technology; National Research Council Canada
Peer reviewedNo
NRC number48761
NPARC number8914077
Export citationExport as RIS
Report a correctionReport a correction
Record identifier1eae05bc-a516-48b4-be8c-d48f7b722065
Record created2009-04-22
Record modified2016-05-09
Bookmark and share
  • Share this page with Facebook (Opens in a new window)
  • Share this page with Twitter (Opens in a new window)
  • Share this page with Google+ (Opens in a new window)
  • Share this page with Delicious (Opens in a new window)