基于互信息的中文术语抽取系统  

Chinese Term Extraction System Based on Mutual Information

作　　者：张锋[1] 许云[1] 侯艳[2] 樊孝忠[1] 

机构地区：[1]北京理工大学计算机科学与工程系,北京100081 [2]广东工业大学计算中心,广东广州510520

出　　处：《计算机应用研究》2005年第5期72-73,77,共3页Application Research of Computers

摘　　要：介绍了一个中文术语自动抽取系统,该系统首先基于互信息计算字串的内部结合强度,从而得到术语候选集;接着从术语候选集中去除基本词,并利用普通词语搭配前缀、后缀信息进一步过滤;最后对术语候选进行词法分析,利用术语的词性构成规则进行判别,得到最终的术语抽取结果。实验结果表明,术语抽取正确率为72. 19%,召回率为77. 98%,F 测量为74. 97%。An automatic Chinese term extraction system based on mutual information is presented.Firstly,the system gets term candidates by calculating internal associative strength of characters string using mutual information. Then, term selection from term candidates is done using basic word dictionary, common collocations suffix, prefix bank and some term POS composing rules. Experiment shows the precision is 72.19%, recall is 77.98% and F measure is 74.97%.

关 键 词：术语抽取 互信息 语料 

分 类 号：TP391[自动化与计算机技术—计算机应用技术]