Constraining a Generative Word Alignment Model with Discriminative Output

Chooi-Ling GOH  Taro WATANABE  Hirofumi YAMAMOTO  Eiichiro SUMITA  

IEICE TRANSACTIONS on Information and Systems   Vol.E93-D   No.7   pp.1976-1983
Publication Date: 2010/07/01
Online ISSN: 1745-1361
DOI: 10.1587/transinf.E93.D.1976
Print ISSN: 0916-8532
Type of Manuscript: PAPER
Category: Natural Language Processing
word alignment,  discriminative model,  generative model,  hybrid,  SMT,  

Full Text: PDF(646.5KB)>>
Buy this Article

We present a method to constrain a statistical generative word alignment model with the output from a discriminative model. The discriminative model is trained using a small set of hand-aligned data that ensures higher precision in alignment. On the other hand, the generative model improves the recall of alignment. By combining these two models, the alignment output becomes more suitable for use in developing a translation model for a phrase-based statistical machine translation (SMT) system. Our experimental results show that the joint alignment model improves the translation performance. The improvement in average of BLEU and METEOR scores is around 1.0-3.9 points.