Abstract:

As an important way of information filtering, collaborative filtering algorithm has attracted more and more attention in the era of big data. However, traditional collaborative filtering algorithm has the problem of low recommendation accuracy due to the serious data sparsity and only considering the scoring similarity between users. This paper proposes an improved collaborative filtering algorithm. Firstly, [K]-means++ algorithm is used to cluster the user attributes, so as to reduce the sparsity of data. Secondly, considering that the user interest will change dynamically with time, this paper introduces the time factor into the traditional scoring similarity. Then, the trust error is introduced into the trust relationship between users, so as to improve the user trust. Finally, the scoring similarity based on the time factor and improved user trust are integrated to improve the calculation accuracy of user similarity. The simulation results on the MovieLens dataset show that the proposed algorithm can effectively improve the prediction accuracy.

Key words: recommendation algorithm, collaborative filtering, [K]-means++, time factor, trust, similarity

摘要：

协同过滤算法作为一种信息筛选的重要方式，在大数据时代下受到越来越多的关注。但传统的协同过滤算法由于面临着严重的数据稀疏性以及只考虑用户间的评分相似性，导致推荐准确率较低。对此，提出了一种改进的协同过滤算法。利用[K]-means++算法对用户属性进行聚类，从而降低数据的稀疏性；考虑到用户兴趣会随时间发生动态变化，在传统的评分相似性中引入时间因素；将信任误差引入到用户间的信任关系中，从而改善用户信任度；将基于时间因素的评分相似性与改进的用户信任度进行融合，从而提高用户相似性的计算精度。在MovieLens数据集上进行仿真实验，结果表明，该算法能有效地提高推荐的预测准确性。

关键词: 推荐算法, 协同过滤, [K]-means++, 时间因素, 信任度, 相似度