中文组织机构名称与简称的识别 - SJTU CS - 上海交通大学.PDF

21 　6 V ol.2 1, No .6 2007 11 JO URNA L OF CHINESE INFO RM A T ION P ROCESSING Nov., 2007 :1003-0077 (2007)06-00 17-05 1 1 2 2 , , , Hans Us koreit (1. 200240;2. ) 　:本文提出了一种基于规则识别中文组织机构全称和简称的方法。 全称的识别首先借助机构后缀 库获得 其右边界, 然后通过规则匹配并借助贝叶斯概率模型加以决策获得其左边界。 简称的识别是在全称的基础上应用 其对应的简称规则实现的。在开放性测试中, 该方法的总体查全率为 85.19%, 查准率为83.03%, F M easure 为 84.10%;简称的查全率为67.18%, 查准率为74.14%。 目前该方法已应用于中文关系的抽取系统。 :计算机应用;中文信息处理;组织机构名称识别;组织机构简称识别;规则匹配;贝叶斯概率模型 :T P39 1　　　　:A Recognition of Chinese Organiz ation Names and Abbreviations 1 1 2 2 SHEN Jia-yi , LI Fang , XU Fei-yu , Hans U s koreit (1.Department of Computer Science and T echnologyp;Shanghai JiaoT ong Univer sity, Shanghai 200240, China; 2.German Research Center for A rtificial Intelligence) Abstract:T his paper pr oposes a method for recog ni ing Chinese org ani ation names and their abbrev iations based on rules.T he right boundary of an org ani ation name is identified w ith the help of the or gani atio n suffix lexico n.T he left boundary is reco gni ed by the optimum rules based o n Bay esian probability model.After idendify ing an org ani a- tion name, we can g et candidate abbr eviations based on abbreviatio n rule s accordingly.In open test, the recall is 85.19%, the precisio n is 83.03%, the F M ea sure is 84.10%for name recog nition, and the recall is 67.18%, the precision is 74.14%for abbreviatio n reco gnition.T his metho d has been applied in the Chinese relation identification sy stem. Key words:co mputer application;Chinese information pro cessing;r eco gnition of Chinese organi atio n name s;rec- o gnition of Chine se organi