关键词: 协同过滤，网络结构，兴趣簇，资源流行度

Abstract: To improve the accuracy of information filtering mechanism and resolve the cold start problem, an information filtering mechanism based on user network architecture was proposed. Here, the social network is treated as a part of the information sharing and dissemination system, like information processing nodes, to help us recognize mobile users' needs and identify more interesting services. The user interest network structure is used to improve the collaborafive filtering. The social network is used to limit the transmission of information. The fingerprints are used to solve the cold start problem. Data analysis shows that this mechanism has better performance.

Key words: Collaborative filtering,Nctwork structure,Interest clustcr,Rcsource popularity