基于知网的词汇集聚分析

词汇集聚提供了文本的结构启示，因而在许多自然语言处理中有极为重要的应用。介绍基于知网分析词汇集聚的方法，首先利用语义网络的形式将知网知识表示出来；然后在文本概念间有引导地建立推理路径；最后基于路径搜索形成体现集聚的词汇链。测试表明，基于知网推理形成的词汇链基本符合人的直觉，准确率较高。

Analysis of Lexical Cohesion Based on HowNet

Lexical cohesion provides an indicator of text structure and therefore,it has been used in many natural language process applications.The paper gives an approach of analyzing lexical cohesion based on HowNet.The knowledge of HowNet is represented in semantic network firstly.And then semantic paths are constructed heuristically between concepts of semantic networks by inference.Lexical chains which express lexical cohesion are found based on semantic paths finally.The results of tests show that the lexical chains got by the introduced method are basically in accord with human's intuition and the precision is reasonable.