Research of Character Level SMT for Patent

摘要：

This paper introduces character level patent translation from Japanese to English. The challenge of this task lies in the segmentation errors caused by the overwhelming OOV (out-of-vocabulary) words exist in the patent texts. In order to solve the problem, we introduce character level SMT. We compare our method with word level SMT model, and find that the two approaches achieve close BLEU scores.

关键词： Statistical machine translation patent translation

作者: Liao Xian-Jun Yang Mu-Yun

作者单位: Library of Academy of Equipment Command & Technology, Beijing 101416, China School of Computer Science and Technology, Harbin Institute of Technology, Harbin 150001, China

会议类型: 国际会议

会议名称: 2010 International Conference on Circuit and Signal Processing(2010年电路与信号处理国际会议 ICCSP 2010)

会议地点: 上海

会议语种:英文

页码: 267-269

在线出版日期: 2010-12-25（万方平台首次上网日期，不代表论文的发表时间）