短文本相似性的改进及其在电商评论推荐中的应用

Improvement of Short Text Similarity and Its Application in E-Commerce Review Recommendation

在常用评论特征的基础上,提出了一种基于搜索引擎(如百度)的文本相似性方法获取评论与产品标题之间的相似性,并作为新的评论特征建立评论推荐模型。实验证明,引入评论与产品相似性特征可明显改进评论推荐机制的有效性,同时文本相似性评价的准确性可以借助搜索引擎得到较大提升。

Based on search engine,a text similarity method is proposed to obtain similarity features between comments and product titles.Combining with other features of comments,a comments recommendation model is established.Experiments show that adding similarity feature can significantly improve the effectiveness of the comment recommendation mechanism.At the same time,the accuracy of text similarity evaluation can be improved by means of search engine.