一种利用统计语义单元识别中文人名的方法  

Method of Identifying Chinese Personal Name Using Statistical Semantic Element

作　　者：余祖波[1] 马建军[1] 

机构地区：[1]大连理工大学计算机科学与工程系,辽宁大连116023

出　　处：《小型微型计算机系统》2008年第12期2339-2343,共5页Journal of Chinese Computer Systems

基　　金：国家高技术研究发展"八六三"计划项目(2006AA012140)资助

摘　　要：从语义语言的角度提出一种利用统计语义单元识别中文人名的方法.在该方法中没有词的概念,一切单位都是语义单元,语义单元有参数和类型等属性.通过语义单元对句子进行语义切分,获得句子的语义单元图,并利用联合概率模型求得语义单元图中概率最大的路径,然后根据人名模式集在该路径上识别人名.初步实验表明,该方法是一种值得探索的新方法.This paper presents a new method of identifying Chinese personal names using statistical semantic element. There is no concept about Chinese word in this method. Anything is semantic element with parameter and type. This method does not employ word segmentation. It divides sentence into some semantic pieces according to semantic elements to get the semantic el- ement graph of the sentence. Then it solves the path with maximum probability of the graph by joint probability model. And it identifies personal names on this path by personal name modes. The experiment results show that this method is a new efficient solution.

关 键 词：自然语言处理 中文人名识别 语义单元 人名模式集 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]