Research of personalized information retrieval system based on multi-agent and user interest model

In order to provide personal information retrieval service for user, this paper introduces a personalized information retrieval system, which can realize individuation and intelligence of information retrieval according to user interest knowledge via multi-agent collaboration. The construction of user interest model based on browsing history record and registration data is discussed in detail, and the system can update user interest model when user´s interest changes. The experimental result indicates that the precision of information retrieval, by use of the personalized information retrieval system based on multi-agent, may be improved 15%-35%. The construction algorithm and update algorithm of user interest model can discover user interest in time, control safely the scale of user interest model, and increase effectively document filtration efficiency.