基于聚类的词表等级关系自动识别研究

Automatic Recognition of Hierarchical Relationship of Thesaurus Based on Word Clustering

词汇等级关系的识别是自动构建叙词表的重点和难点之一。基于相似度的词聚类方法,突破了按字面聚集等级关系词汇的传统做法的局限性,能够深入语义,识别出字面上无此特点的等级关系词汇。介绍了该方法并进行测试,试验结果表明该方法具有一定可行性。

It is difficult but important to realize recognition of hierarchical relationship during automatic thesaurus construction. Breaking through the limit of traditional method , word clustering based on semantic similarity can also recognize hierarchical words without this characters. This article introduce this method, and the test result shows it is feasible.