Abstract: Semantic similarity is one of the basic relations between words; the research on the measure of semantic similarity between Chinese words takes an important role in Chinese language processing applications such as information retrieval, statistical language modeling, etc. In this paper, the semantic similarity based on knowledge graph between words is presented, therefore, knowledge graph, which belongs to the semantic network, is used in the Chinese information processing. The elementary experiment shows that this method is a theoretical approach and effective in practice.

摘要： 语义相似是词汇间的基本关系之一。词汇间语义相似的定量化研究对于信息检索、统计语言模型等的自然语言处理的应用技术具有重要的指导意义。本文提出了基于知识图的词汇间语义相似度计算的方法，把知识图这种属于语义网络范畴的知识表示方法在汉语信息处理中得以应用。初步试验表明，该方法是一种理论基础严整，实践上行之有效的方法。