【摘要】 针对现有的句子相似度计算中仅考虑单个词语的语义信息，而忽略了句子的语法结构信息的问题，提出了一种结合依存句法分析和词语语义相似度的计算方法.使用哈工大社会计算与信息检索研究中心研发的语言技术平台对句子进行句法分析获得依存句法分析树，从中构造包含句子成分、依存关系、词语等多特征信息的依存关系三元组.采用动态加权的方法充分利用词语在知网与同义词词林中的语义信息，通过依存关系三元组体现句子语法结构和词语语义两个层面上的语义信息，提高了相似度计算的合理性，并且扩大了可计算相似度的词语范围.实验表明：该相似度计算方法的准确性相比同类方法有了一定的提高，能更为准确的衡量句子间的相似度.更多还原

【Abstract】 Aiming at the existing sentence similarity calculation that only considers the semantic information of a single word, but ignores the grammatical structure information of the sentence, a calculation method combining dependent syntax analysis and word semantic similarity is proposed. This method uses the language technology platform developed by the Social Computing and Information Retrieval Research Center of Harbin Institute of Technology to perform syntactic analysis on sentences to obtain dependency syntax analysis trees, and construct dependency relation triples containing sentence components, dependency relations, words and other characteristic information. Making full use of the semantic information of the words in HowNet and Tongyici Cilin, and reflecting the semantic information at the two levels of sentence grammatical structure and word semantics through the dependency triples, the dynamic weighting method improves the rationality of similarity calculation and expands the range of words for which similarity can be calculated. Experiments show that the accuracy of the similarity calculation method is improved to a certain extent compared with similar methods, and it can measure the similarity between sentences more accurately.更多还原