Cross Language Dependency Parsing using a Bilingual Lexicon

Proceedings of the 47th Annual Meeting of the ACL and the 4th IJCNLP of the AFNLP, pages 55–63, Suntec, Singapore, 2-7 August 2009. c⃝2009 ACL and AFNLP Cross Language Dependency Parsing using a Bilingual Lexicon∗ Hai Zhao(赵 赵 赵海 海 海)†‡, Yan Song(宋 宋 宋彦 彦 彦)†, Chunyu Kit†, Guodong Zhou‡ †Department of Chinese, Translation and Linguistics City University of Hong Kong 83 Tat Chee Avenue, Kowloon, Hong Kong, China ‡School of Computer Science and Technology Soochow University, Suzhou, China 215006 {haizhao,yansong,ctckit}@cityu.edu.hk, gdzhou@suda.edu.cn Abstract This paper proposes an approach to en- hance dependency parsing in a language by using a translated treebank from an- other language. A simple statistical ma- chine translation method, word-by-word decoding, where not a parallel corpus but a bilingual lexicon is necessary, is adopted for the treebank translation. Using an en- semble method, the key information ex- tracted from word pairs with dependency relations in the translated text is effectively integrated into the parser for the target lan- guage. The proposed method is evaluated in English and Chinese treebanks. It is shown that a translated English treebank helps a Chinese parser obtain a state-of- the-art result. 1 Introduction Although supervised learning methods bring state- of-the-art outcome for dependency parser infer- ring (McDonald et al., 2005; Hall et al., 2007), a large enough data set is often required for specific parsing accuracy according to this type of meth- ods. However, to annotate syntactic structure, ei- ther phrase- or dependency-based, is a costly job. Until now, the largest treebanks1 in various lan- guages for syntax learning are with around one million words (or some other similar units). Lim- ited data stand in the way of further performance enhancement. This is the case for each individual language at least. But, this is not the case as we observe all treebanks in different languages as a whole. For example, of ten treebanks for CoNLL- 2007 shared task, none includes more than 500K ∗The study is partially supported by City University of Hong Kong through the Strategic Research Grant 7002037 and 7002388. The first author is sponsored by a research fel- lowship from CTL, City University of Hong Kong. 1It is a tradition to call an annotated syntactic corpus as treebank in parsing community. tokens, while the sum of tokens from all treebanks is about two million (Nivre et al., 2007). As different human languages or treebanks should share something common, this makes it possible to let dependency parsing in multiple lan- guages be beneficial with each other. In this pa- per, we study how to improve dependency parsing by using (automatically) translated texts attached with transformed dependency information. As a case study, we consider how to enhance a Chinese dependency parser by using a translated English treebank. What our method relies on is not the close relation of the chosen language pair but the similarity of two treebanks, this is the most differ- ent from the previous work. Two main obstacles are supposed to confront in a cross-language dependency parsing …

Content truncated to 500 words