A Data-driven Approach for Cross Transformation Between Mongolian texts

摘要：

　　This paper discusses a data-driven approach to transforming different graphic texts of Mongolian.Using the proposed approach, it is possible to transcribe or translate texts between similar languages such as Mongolian graphic texts used in different regions and countries, as well as the Altaic family languages like Uygur Turkic and Kazakh.The approach has been implemented based on DP (dynamic programming) matching supported by the knowledge-based sequence matching, referred to a multilingual dictionary and a data-driven approach of the target language corpus.Experimental results demonstrate that the proposed method achieves 86.4% transformation accuracy (in F-measure) for the NM (Cyrillic) to the TM (Traditional Mongolian) mainly used in the inner Mongolia, and 91.1% NM to Todo, which is mainly used in Xinjiang areas in China.

关键词： Mongolian texts cross language transformation DP data driven approach

作者: Dawa Yidemucao Muheyat Niyazbek Ayjarken Amantay

作者单位: School of Information Science and Engineering Xinjiang University,Urumqi,China

会议类型: 国际会议

会议名称: 2013 2nd International Conference on Science and Social Research (2013年第二届科学与社会研究国际会议)(ICSSR2013)

会议地点: 北京

会议语种:英文

页码: 375-380

在线出版日期: 2013-07-13（万方平台首次上网日期，不代表论文的发表时间）