Belief Reasoning Recommendation --Mashing up Web Information Fusion and FOAF

This paper mashes up Web Information Fusion and FOAF (Friend of a Friend) to design a belief reasoning recommendation system for user service. Belief reasoning recommendation user service takes into account the spatial and temporal information of the process of user services, and it analysis the relationships between the users and information publisher based on their FOAF profiles. BRRUS uses Markov chain Monte Carlo algorithm to improve the confidence level of cold-start users on virtual social networks who have not registered in BRRUS. This work also proposes an algorithm for particular expert user’s detection from the training datasets according to spatial and temporal queries that users submitted. These theoretical findings are supported by experiments on several test collections and compared with another classic recommendation system Taste, which makes users more satisfactory than other recommendation system.