Bilingual Lexicon Extraction from Comparable Corpora as Metasearch
Résumé
In this article we present a novel way of looking at the problem of automatic acquisition of pairs of translationally equivalent words from comparable corpora. We first present the standard and extended approaches traditionally dedicated to this task. We then reinterpret the extended method, and motivate a novel model to reformulate this approach inspired by the metasearch engines in information retrieval. The empirical results show that performances of our model are always better than the baseline obtained with the extended approach and also competitive with the standard approach.
Origine : Fichiers produits par l'(les) auteur(s)
Loading...