Skip to Main content Skip to Navigation
Theses

Étude d'un modèle computationnel pour la représentation du sens des mots par intégration des relations de contexte

Abstract : In this thesis, we present a theory of concepts and a related computational linguistic model. This non-definitinal theory proposes a view that considers concepts as a Gaussian representation. The term contexonym, formalized version for `contextually related words', is introduced that relates this theory of concepts and computational model. Based on these two principles, our computational model learns automatically contexonyms from a very large untagged corpus. For a given word, the model proposes a set of its contexonyms and organizes them by a hierarchical clustering method. The contexonyms thus obtained reflect encyclopedic knowledge in addition to various linguistic features like word usage or subtle difference between near-synonyms. The results on different tests show that the model could be used for natural language processing (NLP) tasks as well as a dynamic lexical reference.
Complete list of metadata

https://tel.archives-ouvertes.fr/tel-00008384
Contributor : Hyungsuk Ji <>
Submitted on : Monday, February 7, 2005 - 5:38:04 PM
Last modification on : Wednesday, November 20, 2019 - 3:17:24 AM
Long-term archiving on: : Friday, September 14, 2012 - 11:00:33 AM

Identifiers

  • HAL Id : tel-00008384, version 1

Collections

Citation

Hyungsuk Ji. Étude d'un modèle computationnel pour la représentation du sens des mots par intégration des relations de contexte. domain_stic.othe. Institut National Polytechnique de Grenoble - INPG, 2004. Français. ⟨tel-00008384⟩

Share

Metrics

Record views

380

Files downloads

1172