结合用户活跃度的协同过滤推荐算法

Recommendations Based on Collaborative Filtering by User Activity

为了解决协同过滤推荐算法中存在的流行偏置问题,提出一种结合用户活跃度的协同过滤推荐算法(UACF)。该算法考虑用户活跃度对推荐结果的影响,通过对用户活跃度进行聚类分析,针对不同聚类结果中的用户进行分类处理,并引入到相似度计算过程中,以提高相似度计算的可靠性。典型数据集上的对比实验表明该算法能够较好的提高推荐准确率。

In order to solve the problem of popularity bias in recommendation system, we propose a novel collaborative filtering recommendation algorithm, UACF. UACF considers the influence of user activity on the recommended results. It applies cluster analysis algorithm to handle user activity and uses different activity adjustment formula to deal with different clustering results. This method is introduced into the process of similarity calculation to improve the reliability of similarity calculation, experiments on typical data sets show that the al- gorithm can achieve more accurate rating recommendations than the conventional methods.