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

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.