汉英双向时间跟数字命名实体的识别跟翻译系统1anappr资料.pdf

汉英双向时间和数字命名实体的 识别与翻译系统1 翟飞飞 夏睿 周玉 宗成庆 中国科学院自动化所 模式识别国家重点实验室 北京 100190 E-mail: wsknow@ {rxia, yzhou, cqzong}@ 摘要：时间和数字命名实体的识别与翻译是统计机器翻译中语料预处理任务的一个重要步骤，对后续的模 型训练和系统性能有重要的影响。本文在考虑汉英两种语言中时间和数字命名实体构成规律的基础上，将 其进行了类别划分并分别制定了相应的识别和翻译规则，最终实现了一个基于规则方法的时间和数字命名 实体识别翻译工具。经实验测试，该工具具备较高的识别和翻译准确率，提高了统计机器翻译系统的性能。 关键词：时间命名实体，数字命名实体，命名实体识别，命名实体翻译，机器翻译 An Approach to Recognizing and Translating Chinese English Time and Number Named Entities Feifei Zhai, Rui Xia, and Chengqing Zong National Laboratory of Pattern Recognition, Institute of Automation, Chinese Academy of Sciences, Beijing 100190, China E-mail: wsknow@ {rxia, cqzong}@ Abstract : The technique to recognize and translate time and number named entities is an important pre-processing step in statistical machine translation (SMT). It certainly influences the performance of an SMT system. This paper carefully investigates the structural characteristics of time and number named entities in both Chinese and English, divides the time and number named entities into several kinds and formulates the corresponding rules for recognition and translation. Ground on this, a rule-based system has been built to recognize and translate the time and number named entities in both Chinese and English. According to our experiments, the system is shown to have a high accuracy of recognition and translation, and significantly improves the performance of SMT system. Keywords: time named entity, number named entity, named entity recognition, named entity translation, machine translation 1.引言 命名实体(named entity, NE )识别及翻译是信息提取0、问答系统和机器翻译等自然语言 处理任务中重要的基础性问题，在自然语言处理技术走向实用化的过程中占有重要地位。一 1 本论文的研究工作得到国家自然科学基金项目（编号）、国家863 项目（编号：2006AA010108 －4 ）和国家支撑计划项目（编号：2006BAH03B02 ）资助。 般来说，命名实体识别和翻译的任务就是识别并翻译待处理文本中的三大类名称词汇：实体 类、