基于条件随机场的中国人名识别方法  

Recognition of Chinese Personal Name Based on Conditional Random Fields

作　　者：张素香[1] 高国洋[1] 戚银城[1] 

机构地区：[1]华北电力大学电子与通信工程系,河北保定071003

出　　处：《郑州大学学报（理学版）》2009年第2期40-43,共4页Journal of Zhengzhou University:Natural Science Edition

基　　金：华北电力大学博士学位教师科研基金资助项目;编号200812005

摘　　要：充分利用人名的外部特征和内部颗粒特征,提出了一种基于条件随机场的中国人名识别方法.通过建立原子模板和复合模板,将局部特征、关联特征、全局特征以及专家知识相融合,基于条件随机场建立了相应的语言模型,极大地提高了人名识别的准确率和召回率.Making the best of inner features and exterior features,a new method is proposed to recognize Chinese personal name based on conditional random fields.By building atomic pattern and composite patterns,combining local features,global features,related features and expert knowledge, the new linguistics model is put forward based on conditional random fields.And the precision and recall rates of name recognition are greatly improved.

关 键 词：命名实体识别 条件随机场 信息抽取 

分 类 号：TP391.41[自动化与计算机技术—计算机应用技术]