(Utilisation de la similarité sémantique pour l’extraction de lexiques bilingues à partir de corpus comparables) [in French]

This paper presents a new method that aims to improve the results of the standard approach used for bilingual lexicon extraction from specialized comparable corpora. We attempt to solve the problem of context vector word polysemy. Instead of using all the entries of the dictionary to translate a context vector, we only use the words of the lexicon that are more likely to give the best characterization of context vectors in the target language. On two specialised French-English comparable corpora, empirical experimental results show that our method improves the results obtained by the standard approach especially when many words are ambiguous. MOTS-CLES : lexique bilingue, corpus comparable specialise, desambiguisation semantique, WordNet.