摘要： 评分数据的稀疏性影响协同过滤算法的推荐质量。为此，提出一种基于近邻评分填补的混合协同过滤推荐算法。对原始评分矩阵进行全局降维，在低维的主成分空间中计算用户相似性，减少算法复杂度。采用奇异值分解法对近邻评分缺失值进行填补，降低近邻评分的稀疏性。在MovieLens数据集上的实验结果表明，该算法具有较好的推荐效果。

关键词: 推荐系统, 协同过滤, 主成分分析, 近邻评分填补, 稀疏性

Abstract: Data sparsity influences the recommendation quality of collaborative filtering algorithm. To address this problem, a new hybrid collaborative filtering algorithm based on neighbor rating imputation is proposed. The dimensions of original rating matrix are reduced by Principal Component Analysis(PCA), which can reduce the computational complexity. Singular Value Decomposition(SVD) is used to impute missing ratings of the neighbors, which can alleviate the data sparsity. Experiments are carried out on MovieLens dataset, and the results show that the algorithm has higher the recommendation efficiency.

Key words: recommendation system, collaborative filtering, Principal Component Analysis(PCA), neighbor rating imputation, sparsity

中图分类号: 

• TP311