A Chinese-English Organization Name Translation System Using Heuristic Web Mining and Asymmetric Alignment

Proceedings of the 47th Annual Meeting of the ACL and the 4th IJCNLP of the AFNLP, pages 387–395, Suntec, Singapore, 2-7 August 2009. c⃝2009 ACL and AFNLP A Chinese-English Organization Name Translation System Using Heuristic Web Mining and Asymmetric Alignment Fan Yang, Jun Zhao, Kang Liu National Laboratory of Pattern Recognition Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China {fyang,jzhao,kliu}@nlpr.ia.ac.cn Abstract In this paper, we propose a novel system for translating organization names from Chinese to English with the assistance of web resources. Firstly, we adopt a chunking- based segmentation method to improve the segmentation of Chinese organization names which is plagued by the OOV problem. Then a heuristic query construction method is employed to construct an efficient query which can be used to search the bilingual Web pages containing translation equivalents. Finally, we align the Chinese organization name with English sentences using the asymmetric alignment method to find the best English fragment as the translation equivalent. The experimental results show that the proposed method outperforms the baseline statistical machine translation system by 30.42%. 1 Introduction The task of Named Entity (NE) translation is to translate a named entity from the source language to the target language, which plays an important role in machine translation and cross-language information retrieval (CLIR). The organization name (ON) translation is the most difficult subtask in NE translation. The structure of ON is complex and usually nested, including person name, location name and sub-ON etc. For example, the organization name “北京诺基亚通 信 有 限 公 司 (Beijing Nokia Communication Ltd.)” contains a company name (诺基亚/Nokia) and a location name (北京/Beijing). Therefore, the translation of organization names should combine transliteration and translation together. Many previous researchers have tried to solve ON translation problem by building a statistical model or with the assistance of web resources. The performance of ON translation using web knowledge is determined by the solution of the following two problems: � The efficiency of web page searching: how can we find the web pages which contain the translation equivalent when the amount of the returned web pages is limited? � The reliability of the extraction method: how reliably can we extract the translation equivalent from the web pages that we obtained in the searching phase? For solving these two problems, we propose a Chinese-English organization name translation system using heuristic web mining and asymmetric alignment, which has three innovations. 1) Chunking-based segmentation: A Chinese ON is a character sequences, we need to segment it before translation. But the OOV words always make the ON segmentation much more difficult. We adopt a new two-phase method here. First, the Chinese ON is chunked and each chunk is classified into four types. Then, different types of chunks are segmented separately using different strategies. Through chunking the Chinese ON first, the OOVs can be partitioned into one chunk which will not be segmented in the next phase. In this way, the performance of segmentation is improved. 2) Heuristic Query construction: We need to obtain the …

Content truncated to 500 words