Simple window selection strategies for the simplified lesk algorithm for word sense disambiguation

Francisco Viveros-Jiménez, Alexander Gelbukh, Grigori Sidorov

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

6 Scopus citations

Abstract

The Simplified Lesk Algorithm (SLA) is frequently used for word sense disambiguation. It disambiguates by calculating the overlap of a set of dictionary definitions (senses) and the context words. The algorithm is simple and fast, but it has relatively low accuracy. We propose simple strategies for the context window selection that improve the performance of the SLA: (1) constructing the window only with words that have an overlap with some sense of the target word, (2) excluding the target word itself from matching, and (3) avoiding repetitions in the context window. This paper describes the corresponding experiments. Comparison with other more complex knowledge-based algorithms is presented.

Original languageEnglish
Title of host publicationAdvances in Artificial Intelligence and Its Applications - 12th Mexican International Conference on Artificial Intelligence, MICAI 2013, Proceedings
Pages217-227
Number of pages11
EditionPART 1
DOIs
StatePublished - 2013
Event12th Mexican International Conference on Artificial Intelligence, MICAI 2013 - Mexico City, Mexico
Duration: 24 Nov 201330 Nov 2013

Publication series

NameLecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
NumberPART 1
Volume8265 LNAI
ISSN (Print)0302-9743
ISSN (Electronic)1611-3349

Conference

Conference12th Mexican International Conference on Artificial Intelligence, MICAI 2013
Country/TerritoryMexico
CityMexico City
Period24/11/1330/11/13

Fingerprint

Dive into the research topics of 'Simple window selection strategies for the simplified lesk algorithm for word sense disambiguation'. Together they form a unique fingerprint.

Cite this