Selecting influential and trustworthy neighbors for collaborative filtering recommender systems
Ziyang Zhang, Yuhong Liu, Zhigang Jin, Rui Zhang · 2017
Collaborative filtering has become one of the most widely used methods for providing recommendations in various online environments. Its recommendation accuracy highly relies on the selection of appropriate neighbors for the target user/item based on a user-item matrix. However, existing neighbor selection schemes have inevitable inadequacies, especially when handling a very sparse user-item matrix caused by the increasing amount of users and items in recommender systems. To improve the recommendation accuracy, we propose a novel two-layer neighbor selection scheme that improves the quality of neighbor selections by selecting the most influential and trustworthy neighbors. In particular, the proposed scheme consists of two modules: (1) an availability evaluation module and (2) a trust evaluation module, which evaluate users' influence and trustworthiness in providing recommendations. The performance of the proposed scheme is validated through experiments on a real user dataset. Compared to three existing neighbor selection schemes, the proposed scheme achieves higher recommendation accuracy across datasets with different degrees of sparseness.