摘要
论文针对中文短文本特性,综合考虑词性、词序和语义等因素提出了一种新的文本相似度计算方法[1].这种方法是在余弦相似度的基础上计算中文短文本的语义相似度,并通过加权词性、词序权重优化相似度计算结果.实验证明,论文的方法与其他方法相比,准确率和召回率有了显著提高,获得的相似度结果更具说服力.
Abstract
According to the characteristics of Chinese short texts,a method of calculating text similarity is proposed,which combines parts of speech,semantics and word order factors.This method relates the part of speech,meaning and position of words in Chinese short text,and on the basis of cosine formula,through the correlation between the similarity of words of text vectors and the weight of part of speech,this paper improves the method of Chinese short text similarity calculation,and introduces word order similarity to optimize text similarity.The experimental results show that this method has better accuracy and recall than other meth-ods,and is more in line with people's subjective judgment.
维普
万方数据