基于语义依存关系匹配的汉语句子相似度计算

在中文信息处理中,句子相似度计算是一项基础而核心的研究课题,长期以来一直是人们研究的一个热点和难点.句子相似度计算在实际中有着广泛的应用,它的研究状况直接决定着其他一些相关领域的研究进展,如信息检索的相似句子匹配、自动问答的问题匹配与答案抽取等,句子相似度计算都是非常重要的环节.本文提出了一种基于依存关系匹配的句子相似度计算方法,利用词语内在概念和词语相互依存关系计算句子之间的相似度.结果表明,该算法能显著提高返回结果的准确率.

Chinese sentence similarity computing based on semantic dependency matching

Sentence similarity computation is the base and core research topics of Chinese literature information processing. It has a wide range of applications, restricting the development of certain domain, such as sentence matching in information retrieval, answer extraction in question answering. This paper introduces a method of sentence similarity computation based on lexical similarity and sentence dependency matching. Experiments prove that this algorithm improves accuracy of system.