Building wordnets with multi-word expressions from parallel corpora

Alberto Simões, Xavier Gómez Guinovart


In this paper we present a method for enlarging wordnets focusing on multi-word terms and utilising data from parallel corpora. Our approach is validated using the Galician and Portuguese wordnets. The multi-word candidates obtained in this experiment were manually validated, obtaining a 73.2% accuracy for the Galician language and a 75.5% for the Portuguese language.

Texto completo: