A Probabilistic Model for Japanese Zero Pronoun Resolution Integrating Syntactic and Semantic Features.
Kazuhiro Seki, Atsushi Fujii, Tetsuya Ishikawa · 2001
This paper proposes a method to re-solve Japanese zero pronouns by iden-tifying their antecedents. Our method uses a probabilistic model, which is decomposed into syntactic and seman-tic properties. A syntactic model is trained based on corpora annotated with anaphoric relations. However, a se-mantic model is trained based on a large-scale unannotated corpus, so as to counter the data sparseness problem. We also propose the notion of certainty to improve the accuracy of zero pronoun resolution. We show the effectiveness of our proposed method by way of experi-ments. 1