A Localized Prediction Model for Statistical Machine Translation

Proceedings of the 43rd Annual Meeting of the ACL, pages 557–564, Ann Arbor, June 2005. c⃝2005 Association for Computational Linguistics A Localized Prediction Model for Statistical Machine Translation Christoph Tillmann and Tong Zhang IBM T.J. Watson Research Center Yorktown Heights, NY 10598 USA � ctill,tzhang � @us.ibm.com Abstract In this paper, we present a novel training method for a localized phrase-based predic- tion model for statistical machine translation (SMT). The model predicts blocks with orien- tation to handle local phrase re-ordering. We use a maximum likelihood criterion to train a log-linear block bigram model which uses real- valued features (e.g. a language model score) as well as binary features based on the block identities themselves, e.g. block bigram fea- tures. Our training algorithm can easily handle amillions of features. The best system obtains ��� � % improvement over the baseline on a standard Arabic-English translation task. 1 Introduction In this paper, we present a block-based model for statis- tical machine translation. A block is a pair of phrases which are translations of each other. For example, Fig. 1 shows an Arabic-English translation example that uses blocks. During decoding, we view translation as a block� segmentation process, where the input sentence is seg- mented from left to right and the target sentence is gener- ated from bottom to top, one block at a time. A monotone block sequence is generated except for the possibility to swap a pair of neighbor blocks. We use an orientation model similar to the lexicalized block re-ordering model in (Tillmann, 2004; Och et al., 2004): to generate a block � with orientation relative to its predecessor block � . During decoding, we compute the probability  �   � of a block sequence �  with orientation   as a product of block bigram probabilities:  �            �    �        (1) "! #%$"&('")%$"*+$ ,.- /.0 '"1 $ 2 '"34 0 '")%$"* 56*+3'"$ 0 - 7 0 8 7:9 3 7 1 9 7 0 ; 3 &< = 7 0 7 * 3 7 < 0 < = 1 ) 1> ? ' - 3*+4('"@+$ 0 0 & ) 7 ) < 7 0 ACB 7 0 7 0 B 2 < 7 "D "E GF Figure 1: An Arabic-English block translation example, where the Arabic words are romanized. The following orientation sequence is generated: IHKJ  ML HKN  MO H J  MP HKQ . where �  is a block and SRUT N  eft   Q  ight V J  eutral 6W is a three-valued orientation component linked to the block �  (the orientation   of the predecessor block is currently ignored.). Here, the block sequence with ori- entation  �X      is generated under the restriction that the concatenated source phrases of the blocks �  yield the input sentence. In …

Content truncated to 500 words