基于奇异值分解的个性化评论推荐

Singular Value Decomposition-Based Personalized Review Recommendation

针对如何让消费者在海量评论中快速找到自己感兴趣的评论,该文提出了一个基于奇异值分解的个性化评论推荐系统Rev Rec Sys。该方法首先构建了用户-特征矩阵和评论-特征矩阵;然后利用矩阵分解技术把这两个矩阵压缩到隐因子向量空间;最后通过匹配用户的隐因子向量空间和评论的隐因子向量空间实现评论推荐。通过实验,验证了Rev Rec Sys相比现有的方法,可以获得更好的推荐效果。

With the boom of reviews available in e-commerce websites, it is time-consuming for customers to find their interested reviews. Motivated by this situation, we propose a framework named Rev Rec Sys based on singular value decomposition（SVD） for personalized review recommender systems. Our framework first constructs user-feature matrix and review-feature matrix, then it adopts matrix factorization techniques to compress these two matrices into latent factors, finally it matches a user＇s latent factor vector space and a review＇s latent factor vector space to achieve review recommendation. To evaluate the proposed framework, we conduct experiments on a real-life data set. The experimental results report that our method can achieve a better performance than the baseline methods.