Research And Design Of Personalized Recommendation System Based On Agent

When faced the complex and huge internet , most users feel hard to search information they want, and always lost their ways because of the overwhelmed information and lack of purposes. Base on this situation , the Personalized Recommendation System emerge as the times require and develops quickly .The Personalized Recommendation System , base on the current user's accessing page and the access habits of other users , decide which group the current user belong to ,then predict the next pages that users may be interested in and send to the user.Base on a further study of general process and technology of WEB LOG mining ,this paper focused on the research of WEBLOG Mining Algorithm . And at last this paper propose a matrix algorithm based on interest measure which improve and use the interest measure to fill the matrix instead of the custom 1,0 algorithm . The interest measure combined reading time with reading times which can most stand the user's interest . The results is not only efficient as the 1,0 algorithm but also more similar to the real user visit model .At the meantime , this paper deeply researched on the Agent concept ,technology frame , the communication between Agents , JADE workshop and the programming of Agent. And because of the communication property ,autonomy, cooperation property ,the mining system architecture base on Agent is proposed . The functions and the way of transfer of Agents are given in detail ,the solution of communication problem in MAS system is discussed base on XML and KQML communication language , the independent Agent programming in the real development process is described , and at last both the implementation interface and the results of user cluster and page cluster are shown .