Optimize the WEB personalized recommender model using market mechanism

Based on the analysis of the current hybrid recommender systems, this paper proposes a new recommender system framework to overcome the disadvantage of these hybridization technologies. In the new system, various Web personalized recommending methods are integrated into a market model. It is capable of producing recommendations for the unregistered users. And a finely reasonable auction process is designed to do the market´s job and the key procedures are detailed. This model is able to serve unregistered users with high-quality recommendations continuously under the effect of the market´s positive feedback. The preliminary experiments show the feasibility and effectiveness of the optimization approach.