摘要： 针对在用户评分数据极端稀疏环境下传统协同过滤推荐算法存在的弊端，从提高邻居用户识别准确性出发，对传统相似性度量方法进行改进，在此基础上提出一种基于相关均值的推荐算法。实验结果表明，该算法能增强邻居用户在推荐中的影响力，有效提高推荐精度，改善推荐质量。

关键词: 协同过滤, 相似性度量, 相关均值, 平均绝对偏差

Abstract: According to the disadvantage of the traditional collaborative algorithm while the user rating data extremely sparse, this paper proposes a novel similarity measure method and a recommendation algorithm based on Correlation Mean(CM). Experimental results show it can enhance the neighbor’s influence in the course of recommendation, and improve the accuracy and the quality of recommendation system effectively.

Key words: collaborative filter, similarity measure, Correlation Mean(CM), Mean Absolute Error(MAE)

中图分类号: 

• TP311.52