Improving fast_align by Reordering

Proceedings of the 2015 Conference on Empirical Methods in Natural Language Processing, pages 1034–1039, Lisbon, Portugal, 17-21 September 2015. c⃝2015 Association for Computational Linguistics. Improving fast align by Reordering Chenchen Ding, Masao Utiyama, Eiichiro Sumita Multilingual Translation Laboratory National Institute of Information and Communications Technology 3-5 Hikaridai, Seikacho, Sorakugun, Kyoto, 619-0289, Japan {chenchen.ding, mutiyama, eiichiro.sumita}@nict.go.jp Abstract fast align is a simple, fast, and effi- cient approach for word alignment based on the IBM model 2. fast align per- forms well for language pairs with rel- atively similar word orders; however, it does not perform well for language pairs with drastically different word orders. We propose a segmenting-reversing reorder- ing process to solve this problem by al- ternately applying fast align and re- ordering source sentences during train- ing. Experimental results with Japanese- English translation demonstrate that the proposed approach improves the per- formance of fast align significantly without the loss of efficiency. Experiments using other languages are also reported. 1 Introduction Aligning words in a parallel corpus is a basic task for almost all state-of-the-art statistical machine translation (SMT) systems. Word alignment is used to extract translation rules in various way, such as the phrase pairs used in a phrase-based (PB) SMT system (Koehn et al., 2003), the hier- archical rules used in a HIERO system (Chiang, 2007), and the sophisticated translation templates used in tree-based SMT systems (Liu et al., 2006). Among different approaches, GIZA++1 (Och and Ney, 2003), which is based on the IBM translation models, is the most widely used word alignment tool. Other well-known tools are the BerkeleyAligner2, Nile3 (Riesa et al., 2011), and pialign4 (Neubig et al., 2011). 1http://www.statmt.org/moses/giza/ GIZA++.html 2https://code.google.com/p/ berkeleyaligner/ 3http://jasonriesa.github.io/nile/ 4http://www.phontron.com/pialign/ fast align5 (Dyer et al., 2013) is a recently proposed word alignment approach based on the reparameterization of the IBM model 2, which is usually referred to as a zero-order alignment model (Och and Ney, 2003). Taking advantage of the simplicity of the IBM model 2, fast align introduces a “tension” parameter to model the overall accordance of word orders and an effi- cient parameter re-estimation algorithm is devised. It has been reported that the fast align ap- proach is more than 10 times faster than baseline GIZA++, with comparable results in end-to-end French-, Chinese-, and Arabic-to-English transla- tion experiments. However, the simplicity of the IBM model 2 also leads to a limitation. As demonstrated in this study, fast align does not perform well when applied to language pairs with drastically different word orders, e.g., Japanese and English. The problem is because of the IBM model 2’s in- trinsic inability to handle complex distortions. In this study, we propose a simple and efficient re- ordering approach to improve the fast align’s performance in such situations, referred to as segmenting-reversing (seg rev). Our motivation is to apply a rough but robust reordering to make the source and target sentences have more simi- lar word orders, where fast align can show its power. Specifically, seg rev first segments a source-target sentence pair into …

Content truncated to 500 words