基于稀疏聚类和信任度的协同过滤算法

Collaborative Filtering Algorithm Based on Sparse Clustering and Trust Degree

针对数据稀疏性问题,提出一种基于稀疏子空间聚类和预测评分的协同过滤算法。利用稀疏子空间聚类对用户评分矩阵进行聚类,可以保留更多有用信息。考虑用户评分尺度和用户之间的可信度问题,提出融合信任度的概念,通过计算用户间的信任度,最终使用用户间的信任度与相似度的结合作为新的权重进行推荐。

In order to solve the problem of data sparsity, a collaborative filtering algorithm based on sparse subspace clustering and prediction score is proposed. The sparse subspace clustering is used to cluster the user score matrix, and more useful information can be retained. Considering the user rating scale and the credibility between users, the concept of fusion trust degree is proposed. By calculating the degree of trust among users, the trust degree and similarity between users are finally used as a new weight to recommend.