A hierarchy weighting similarity measure to improve user-based collaborative filtering algorithm
Wenqiang Li, Hongji Xu, Mingyang Ji, Zhengzheng Xu, Haiteng Fang · 2016
The aim of recommender systems is to help users to find items that they should be interested in from over-load information by analyzing historical information about the users to establish the interesting model. In this area, user-based collaborative filtering recommendation algorithm is one of the most popular techniques, especially in blog and news recommendation area. However, due to the poor distinction of similarity between users, the effectiveness of existing recommendation methods could decrease greatly. In this paper, we propose and analyze a hierarchy weighting similarity measure which weights the similarity at different levels. Extensive experiments are conducted on the publicly available datasets. Experimental results indicate that the proposed method shows a significant improvement over existing approaches in rating prediction.