一种基于贝叶斯分类与机读词典的多义词排歧方法  

Method of word sense disambiguation based on bayes and machine readable dictionary

作　　者：谈文蓉[1] 符红光[2] 刘莉[1] 杨宪泽[1] 

机构地区：[1]西南民族大学计算机科学与技术学院,四川成都610041 [2]中国科学院成都计算机应用研究所,四川成都610041

出　　处：《计算机应用》2006年第6期1389-1391,1395,共4页journal of Computer Applications

基　　金：四川省重点科技攻关项目(05SG022-016)

摘　　要：一词多义是自然语言中普遍存在的现象,词义排歧的成功率是衡量机器翻译、信息检索、文本分类等自然语言处理软件性能的重要指标。提出了一种基于贝叶斯分类与机读词典的多义词排歧方法,通过小规模语料库的训练和歧义词在机读词典中的语义定义来完成歧义的消除。实验表明:基于贝叶斯分类与机读词典的多义词排歧算法在标注语料库规模受限的情况下,能取得较高的排歧准确率。multl-senses of a word are widespread phenomenon in the natural language. The accuracy rate of sense ambiguation is the most important target of a software on the fields of machine translation, information indexing and text sorting. A method based on the bayes and machine readable dictionary was proposed, which could disambiguate by the training of a small-scale corpus and the definition of semantic in machine dictionary. The experimental results show that it has a high accuracy rate of word sense ambiguation when the scale of markup corpus has been limited.

关 键 词：词义排歧 语料库 机读词典 自然语言处理 

分 类 号：TP181[自动化与计算机技术—控制理论与控制工程]