The Design Of The Personalized Publication Information Service System Based On Agent

With the rapid development of Internet technology, the computer terminal can easily have a flood of information resources. The paths that people access to information has increased, but also allow the majority of users feel powerless. When users search for information on the network, they often will be "Trek," "cognitive overload" and so on, because of overload purposeless information. In addition, with the accumulation of information resources, huge data from a single server or storage has also brought a lot of inconvenience for search and management.To this end, this paper first studied the traditional Content-based Filtering and Collaborative Filtering technology, improving the traditional Collaborative Filtering techniques by constructing classification evaluation matrix and giving a new formula for collaborative filtering recommendation. Combining Content-based Filtering recommendation techniques, it proposed a distributed mixed recommendation model that involved the advantages of Content-based Filtering and Collaborative Filtering, and recommendation by this model has been proved effective and scientific. Secondly, it proposed a WEB mining model based on Agent which is using a distributed architecture and designed the features of the entire Agent in detail. Through the studying Agent module, it can continuously update knowledge base. It took the distributing mixed recommendation model as the core algorithm of information filtering, so as to recommend the most interesting projects for users. XML-based solution is proposed to meet the information interact and interoperability between multi-agents. Once more, the distributed mixed recommendation model and the WEB mining model based on Agent are applied to publications service system. Designing the two levels of database was used to support Content-based Filtering and Collaborative Filtering module respectively. Taking full advantage of the distributed architecture and Agent technology, it can accomplish distributed intelligent search and personalized recommendations across the site, it is to say, for the same keyword query, different kinds of users will search for a different results in order to meet the needs of different types of users. Finally, the system interface map implementation was also to be showed in the paper.At last part of the paper, all of researches and work in the paper are summarized, and many aspects of improvements are proposed. In the same time, the further research directions of the field of Agent technology in personalized recommendations and distributed intelligent search areas are prospected.