A Bayesian approach combining surface clues and linguistic knowledge: Application to the anaphora resolution problem

Davy Weissenbacher (1) , Adeline Nazarenko (1)

1 LIPN - Laboratoire d'Informatique de Paris-Nord

In NLP, A traditional distinction opposes the linguistically-based systems and the knowledge-poor ones which mainly rely on surface clues. Each approach has its drawbacks and its advantages. In this paper, we propose a new method which is based on Bayes Networks and allows to combine both types of information. As a case study, we focus on the specific task of pronominal anaphora resolution which is known as a difficult NLP problem. We show that our bayesian system performs better than state-of-the art anaphora resolution ones.

Domaines: Traitement du texte et du document

Davy Weissenbacher, Adeline Nazarenko. A Bayesian approach combining surface clues and linguistic knowledge: Application to the anaphora resolution problem. Recent Advances in Natural Language Processing, Sep 2007, Borovets, Bulgaria. 7 p. (édition électronique). ⟨hal-00162114⟩