摘要： 提出一种对协同推荐的pLSA模型进行修正的算法,在推荐的过程中通过逐步修正PLSA模型中用户的兴趣模型,来适应用户兴趣的变化。通过实验证明该算法能够取得更好的推荐准确度。

关键词: 推荐系统, Probabilistic latent semantic analysis(pLAS), 协同过滤

Abstract: This paper proposes an approach to revise the pLSA model after each recommendation made.Based on the observation that single user interests changes fast,but the community interests remain comparably constant,it can keep parameters that represent the community interests as constant,while those represent the user’s interests as variables.After each recommendation made,the paper revises the variables for a single user with maximum estimation.Experimental evaluation against the MovieLens data set shows that proposed method improves the recommending accuracy further beyond the original pLSA model at a relatively low computation cost.

中图分类号: 

• TP393.09