Abstract: This paper describes a technique for extracting idioms from text. The technique works by finding patterns such as “thrills and spills,” whose reversals (such as “spills and thrills”) are never encountered. This method collects not only idioms, but also many phrases that exhibit a strong tendency to occur in one particular order, due apparently to underlying semantic issues. These include hierarchical relationships, gender differences, temporal ordering, and prototype-variant effects.
Appeared in ACL2005 Workshop on Deep Lexical Acquisition, Ann Arbor, Michigan, June 30th, 2005
Download File: Download PDF (2MB)
Authors
Dr. Dominic Widdows, Beate Dorow (Institute for Natural Language Processing, University of Stuttgart)
Publish Date
Mar 2008