Title

Evaluating Variants of the Lesk Approach for Disambiguating Words

Author(s)

Florentina Vasilescu, Philippe Langlais, Guy Lapalme

RALI/IRO, Université de Montréal, CP 6128, succursale Centre-ville, {vasilesf, felipe, lapalme}@iro.umontreal.ca

Session

O11-EW

Abstract

This paper presents a detailed analysis of the factors determining the performance of Lesk-based word sense disambiguation methods. We conducted a series of experiments on the original Lesk algorithm, adapted to WORDNET, and on some variants. These methods were evaluated on the test corpus from SENSEVAL2, English All Words, and on excerpts from SEMCOR. We designed a fine grain analysis of the answers provided by each variant in order to better understand the algorithms than by the mere precision and recall figures.

Keyword(s)

Word Sense Disambiguation, Lesk algorithm, Naïve Bayes, WORDNET

Language(s) English
Full Paper

219.pdf