IEICE TRANSACTIONS on Information
Constraining a Generative Word Alignment Model with Discriminative Output
Summary :
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.
- Publication
- IEICE TRANSACTIONS on Information Vol.E93-D No.7 pp.1976-1983
- Publication Date
- 2010/07/01
- Publicized
- Online ISSN
- 1745-1361
- DOI
- 10.1587/transinf.E93.D.1976
- Type of Manuscript
- PAPER
- Category
- Natural Language Processing
Authors
Keyword
Latest Issue
Copyrights notice of machine-translated contents
The copyright of the original papers published on this site belongs to IEICE. Unauthorized use of the original or translated papers is prohibited. See IEICE Provisions on Copyright for details.
Cite this
Copy
Chooi-Ling GOH, Taro WATANABE, Hirofumi YAMAMOTO, Eiichiro SUMITA, "Constraining a Generative Word Alignment Model with Discriminative Output" in IEICE TRANSACTIONS on Information,
vol. E93-D, no. 7, pp. 1976-1983, July 2010, doi: 10.1587/transinf.E93.D.1976.
Abstract: 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.
URL: https://global.ieice.org/en_transactions/information/10.1587/transinf.E93.D.1976/_p
Copy
@ARTICLE{e93-d_7_1976,
author={Chooi-Ling GOH, Taro WATANABE, Hirofumi YAMAMOTO, Eiichiro SUMITA, },
journal={IEICE TRANSACTIONS on Information},
title={Constraining a Generative Word Alignment Model with Discriminative Output},
year={2010},
volume={E93-D},
number={7},
pages={1976-1983},
abstract={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.},
keywords={},
doi={10.1587/transinf.E93.D.1976},
ISSN={1745-1361},
month={July},}
Copy
TY - JOUR
TI - Constraining a Generative Word Alignment Model with Discriminative Output
T2 - IEICE TRANSACTIONS on Information
SP - 1976
EP - 1983
AU - Chooi-Ling GOH
AU - Taro WATANABE
AU - Hirofumi YAMAMOTO
AU - Eiichiro SUMITA
PY - 2010
DO - 10.1587/transinf.E93.D.1976
JO - IEICE TRANSACTIONS on Information
SN - 1745-1361
VL - E93-D
IS - 7
JA - IEICE TRANSACTIONS on Information
Y1 - July 2010
AB - 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.
ER -
English
