Evaluation and Comparison of multilingual fusion strategies for similarity-based Word Sense Disambiguation
Résumé
In this article, we investigate the e↵ects on the quality ofthe disambiguation of exploiting multilingual features with a similarity-based WSD system based on an Ant Colony Algorithm. We consideredfeatures from one, two, three or four languages in order to quantify theimprovement brought by using features from additional languages. Us-ing BabelNet as a multilingual resource, we considered three data fu-sion strategies: an early fusion strategy, and two late fusion strategies(majority vote and weighted majority vote). We found that the early fu-sion approach did not produce any significant improvements while votingstrategies adding features from more languages led to an increase in thequality of the disambiguation of up to 2.84%. Furthermore, a simplemajority vote led to better results than the weighted variant.
Domaines
Intelligence artificielle [cs.AI]
Origine : Fichiers produits par l'(les) auteur(s)