Bilingual Cluster Based Models for Statistical Machine Translation

Proceedings of the 2007 Joint Conference on Empirical Methods in Natural Language Processing and Computational Natural Language Learning, pp. 514–523, Prague, June 2007. c⃝2007 Association for Computational Linguistics Bilingual Cluster Based Models for Statistical Machine Translation Hirofumi Yamamoto National Institute of Information and Communications Technology / 2-2-2 Hikaridai Seika-cho Soraku-gun Kyoto Japan and ATR Spoken Language Communication Research Labs. hirofumi.yamamoto@nict.go.jp Eiichiro Sumita National Institute of Information and Communications Technology / 2-2-2 Hikaridai Seika-cho Soraku-gun Kyoto Japan and ATR Spoken Language Communication Research Labs. eiichiro.sumita@nict.go.jp Abstract We propose a domain specific model for statistical machine translation. It is well- known that domain specific language mod- els perform well in automatic speech recog- nition. We show that domain specific lan- guage and translation models also benefit statistical machine translation. However, there are two problems with using domain specific models. The first is the data sparse- ness problem. We employ an adaptation technique to overcome this problem. The second issue is domain prediction. In or- der to perform adaptation, the domain must be provided, however in many cases, the domain is not known or changes dynami- cally. For these cases, not only the trans- lation target sentence but also the domain must be predicted. This paper focuses on the domain prediction problem for statisti- cal machine translation. In the proposed method, a bilingual training corpus, is au- tomatically clustered into sub-corpora. Each sub-corpus is deemed to be a domain. The domain of a source sentence is predicted by using its similarity to the sub-corpora. The predicted domain (sub-corpus) specific lan- guage and translation models are then used for the translation decoding. This approach gave an improvement of 2.7 in BLEU (Pa- pineni et al., 2002) score on the IWSLT05 Japanese to English evaluation corpus (im- proving the score from 52.4 to 55.1). This is a substantial gain and indicates the validity of the proposed bilingual cluster based mod- els. 1 Introduction Statistical models, such as n-gram models, are widely used in natural language processing, for ex- ample in speech recognition and statistical machine translation (SMT). The performance of a statistical model has been shown to improve when domain spe- cific models are used, since similarity of statistical characteristics between model and target is higher. For utilize of domain specific models, a training data sparseness and target domain estimation prob- lems must be resolved. In this paper, we try to es- timate target domain sentence by sentence, consid- ering cases where the domain changes dynamically. After sentence by sentence domain estimation, do- main specific models are used for translation using the adaptation technique(Seymore et al., 1997). In order to train a classifier to predict the domain, we used an unsupervised clustering technique on an unlabelled bilingual training corpus. We regarded each cluster (sub-corpus) as a domain. Prior to trans- lation, the domain of the source sentence is first pre- dicted and this prediction is then used for model se- lection. The most similar sub-corpus to the transla- tion source sentence is used to represent …

Content truncated to 500 words