基于用户和项目组合的协同过滤推荐算法

An Improved Collaborative Filtering Algorithm Based on Combining User with Item

协同过滤算法是电子商务系统中一种重要的个性化推荐技术之一。但是随着电子商务规模的扩大，评分矩阵的稀疏性问题严重的影响了协同过滤算法的推荐效果。该文通过分析并研究了传统的协同过滤算法的不足，提出了一种新的基于用户和项目组合的协同过滤算法，在对稀疏矩阵进行填充时，不仅考虑到了项目之间的相关性，还考虑到了用户之间的相关性，然后在此基础上，构造虚拟的评分矩阵，最后再进行综合推荐。实验结果表明，在评分矩阵极其稀疏的情况下，该算法能有效的提高预测精度。

Collaborative Filtering Algorithm is one of the important personalized recommendation technologies, however, as the increasing scale of the ecommerce, the rating matrix is quite sparse. Thus the quality of the approach is seriously deceased. This paper proposes a new improved approach that based on combining user with item by analyzing the deficiency of the traditional algorithm. In this algorithm, we take the information of the users and items into account while predicting the missing data, then build a virtual matrix to recommend the items. The experimental results show that this new approach can efficiently improve the quality of the recommendation.