融合用户信任的协同过滤推荐算法

Collaborative Filtering Recommendation Algorithm Combined with User Trust

针对协同过滤推荐算法中存在的准确率较低、数据稀疏等问题,提出基于用户信任的协同过滤推荐算法,算法包含计算用户之间评分信任度和偏好信任度2个部分.对于用户项目评分矩阵中的用户间共同评分项目,综合考虑共同评分项目的数量以及其在所有评价项目中所占的比例,并与用户评分相似度结合,建立非对称的评分信任矩阵,计算用户评分信任度.对于非共同评分项目,利用项目自身的标签信息以及用户评分权重,计算用户偏好信任度.然后算法将评分信任度和偏好信任度线性加权融合.最后在真实数据集上与相关算法进行实验对比,实验结果表明,提出的算法在推荐的准确率以及评分预测上取得了较好的效果.

To address problems such as data sparsity and low accuracy commonly found in the recommendation system, this paper proposes a collaborative filtering recommendation algorithm combined with user trust. The algorithm includes 2 parts, which are user＇s score trust and preference trust. In view of the common score items of user item score matrix, the algorithm considers the number of the common score items and the proportion of the evaluation items, and combined with the score similarity between users, establish an asymmetric trust relationship matrix,and calculate the user＇s score trust. For the non common score project,we use the attribute information of items and the scoring weight to calculate the user＇s preference trust. Then, a new recommendation algorithm based on user trust is proposed. The results show that the proposed algorithm can effectively improve the accuracy of recommendation.