FOAF-based distributed desktop system for programs recommendation

Recent research has shown that simple network user recognition is not enough for FOAF (Friend of a Friend) user analysis, and meanwhile more and more researchers focus on user interest analysis and products recommendation. Extremely dispersed mapping information of customer relationship and space storage dispersion make it a challenge to effectively mine and analyze the FOAF information. In this paper, we apply FOAF and DBlink (database links) to recognizing the users' identity and to mining the interest of users so as to recommend latest online TV and movie programs for users from various online programs web sites and whose FOAF flies stored in different databases and different computers. This essay focuses on applying FOAF to a latest online TV program recommendation system for a particular user from various online video web sites he/she has registered in. The article describes the approach to such services based on HMM (Hidden Markov Model). For the protection of user privacy, the system is used as local-service desktop model. We conduct experiments to illustrate users' high degree of satisfaction to our techniques.