SMT Helps Bitext Dependency Parsing

Proceedings of the 2011 Conference on Empirical Methods in Natural Language Processing, pages 73–83, Edinburgh, Scotland, UK, July 27–31, 2011. c⃝2011 Association for Computational Linguistics SMT Helps Bitext Dependency Parsing Wenliang Chen†‡, Jun’ichi Kazama‡, Min Zhang†, Yoshimasa Tsuruoka∗‡, Yujie Zhang⋆‡, Yiou Wang‡, Kentaro Torisawa‡ and Haizhou Li† †Human Language Technology, Institute for Infocomm Research, Singapore ‡National Institute of Information and Communications Technology (NICT), Japan ∗School of Information Science, JAIST, Japan ⋆Beijing Jiaotong University, China {wechen, mzhang, hli}@i2r.a-star.edu.sg {kazama, torisawa, yujie, wangyiou}@nict.go.jp tsuruoka@jaist.ac.jp Abstract We propose a method to improve the accuracy of parsing bilingual texts (bitexts) with the help of statistical machine translation (SMT) systems. Previous bitext parsing methods use human-annotated bilingual treebanks that are hard to obtain. Instead, our approach uses an auto-generated bilingual treebank to produce bilingual constraints. However, because the auto-generated bilingual treebank contains er- rors, the bilingual constraints are noisy. To overcome this problem, we use large-scale unannotated data to verify the constraints and design a set of effective bilingual features for parsing models based on the verified results. The experimental results show that our new parsers significantly outperform state-of-the- art baselines. Moreover, our approach is still able to provide improvement when we use a larger monolingual treebank that results in a much stronger baseline. Especially notable is that our approach can be used in a purely monolingual setting with the help of SMT. 1 Introduction Recently there have been several studies aiming to improve the performance of parsing bilingual texts (bitexts) (Smith and Smith, 2004; Burkett and Klein, 2008; Huang et al., 2009; Zhao et al., 2009; Chen et al., 2010). In bitext parsing, we can use the in- formation based on “bilingual constraints” (Burkett and Klein, 2008), which do not exist in monolingual sentences. More accurate bitext parsing results can be effectively used in the training of syntax-based machine translation systems (Liu and Huang, 2010). Most previous studies rely on bilingual treebanks to provide bilingual constraints for bitext parsing. Burkett and Klein (2008) proposed joint models on bitexts to improve the performance on either or both sides. Their method uses bilingual treebanks that have human-annotated tree structures on both sides. Huang et al. (2009) presented a method to train a source-language parser by using the reordering in- formation on words between the sentences on two sides. It uses another type of bilingual treebanks that have tree structures on the source sentences and their human-translated sentences. Chen et al. (2010) also used bilingual treebanks and made use of tree structures on the target side. However, the bilingual treebanks are hard to obtain, partly because of the high cost of human translation. Thus, in their experi- ments, they applied their methods to a small data set, the manually translated portion of the Chinese Tree- bank (CTB) which contains only about 3,000 sen- tences. On the other hand, many large-scale mono- lingual treebanks exist, such as the Penn English Treebank (PTB) (Marcus et al., 1993) (about 40,000 sentences in Version 3) and the latest version of CTB …

Content truncated to 500 words