Personalized query evaluation in ring-based P2P networks

Personalized query evaluation which returns top-k results according to users' individual interests was once confined mainly to the web literature, while now it is increasingly becoming a factor in many other search systems, such as peer-to-peer systems, cloud computing systems etc. In this paper, we propose a novel method for efficient personalized query processing in P2P systems. The problem has never been discussed and remains imperative to be solved. By effectively calculating a corresponding search range on-the-fly which contains the dedicated top-k results based on the estimated multiscale data density in the current system, a personalized query is transformed into a special range query for efficient processing at run time. Singular value decomposition of the personalization matrix is deployed to simplify the process of range computation. Search request is then multicasted within the desired range, which optimizes the communication cost when fetching those results in the P2P network. Algorithms on multiscale density estimation, range computation and multicast routing are given as well as the theoretical analysis. Our extensive performance study confirms the effectiveness and efficiency of our method.