Paper: Using Distributional Similarity To Identify Individual Verb Choice

ACL ID W06-1406
Title Using Distributional Similarity To Identify Individual Verb Choice
Venue International Conference on Natural Language Generation
Session Main Conference
Year 2006
Authors
  • Jing Lin (University of Aberdeen, Aberdeen UK)

Human text is characterised by the indi- vidual lexical choices of a specific au- thor. Significant variations exist be- tween authors. In contrast, natural lan- guage generation systems normally pro- duce uniform texts. In this paper we apply distributional similarity measures to help verb choice in a natural lan- guage generation system which tries to generate text similar to individual au- thor. By using a distributional sim- ilarity (DS) measure on corpora col- lected from a recipe domain, we get the most likely verbs for individual au- thors. The accuracy of matching verb pairs produced by distributional similar- ityishigherthanusingthesynonymout- puts of verbs from WordNet. Further- more, the combination of the two meth- ods provides the best accuracy.