Using WordNet^?(TM) to Disambiguate Word Sences for Text Retrieval

Date	Place	Author	Keyword
1993	SIGIR	Ellen M. Voorhees	Indexing word sense classes.

Summary

This paper uses Princeton University's WordNet^? lexical system to index words senses, as opposed to word stems. The results show that this technique causes a degradation in retrieval effectiveness.

Background and Motivation

This research tries to deal with the issues of polysemy and synonymy inherent in natural language. WordNet^? seems like a natural choice to get around these issues as it includes "IS-A" relations between words.

Contribution

* Demonstrates that word sense disambiguation shows no improvement over stem-based indexing when used to index a collection.

Reference

• Original Paper

-- HenryFeild - 16 Oct 2007
to top