Abstract: In order to improve the accuracy of sentence similarity from structures similarity, this paper proposes a similarity measure method of Chinese sentence structures. It performs the optimal matching between the Part Of Speech(POS) sequences and POS dependency of two compared sentences. Experimental results show that the new method works well and it is more reasonable than the other methods.

Key words: sentence structures, Part Of Speech(POS) dependency, structures similarity, natural language processing

摘要： 为提高句子相似度的准确率，从结构相似度出发，提出基于词性及词性依存关系的句子结构相似度计算方法。该方法从正向和逆向比较句子的词性序列，获得2个句子词性及词性依存关系的最优匹配，从而计算句子结构相似度。实验结果表明，该方法能使句子结构相似度计算更合理。

关键词: 句子结构, 词性依存, 结构相似度, 自然语言处理

CLC Number: 

• TP391