中文短文本语法语义相似度算法

A Chinese Short Text Similarity Algorithm Based on Semantic and Syntax

通过分析中文短文本的特征,提出了一种基于语法语义的短文本相似度算法.该算法结合中文语句语义的相似性以及语句语法的相似性,即计算具有相同句法结构的短文本的相似度以及考虑语句词组顺序对相似度的贡献,对中文短文本相似度进行计算.实验表明,本文提出的算法在中文短文本相似度计算结果上更加接近人们的主观判断并且拥有比较好的精确率与召回率.

A short text algorithm based on semantic and syntax by analyzing the characteristics of Chinese short text was proposed. The algorithm combines semantic similarity with the similarity of sentence syntax in Chinese short text, that is to say, to calculate the similarity of short text with the same syntax structure and consider the contribution with sentence words order to similarity. The experiments show that the proposed algorithm is closer to people＇s judgment and gets a better precision rate and recall rate in the aspect of Chinese short text similarity calculation results.