A Study On Agent Based Information Recommender System

Recommender System is developed to help people obtain the just needed information from large-scale information collections Content-based filtering and collaborative filtering are the most common information filtering technologies in recommender system, but each of them has its own advantages as well as disadvantages. Information recommender system based on traditional Client/Serve architecture doesn't meet the need of the changing application in network. Agent paradigm has evolved as a promising distributed computing paradigm in recent years characterized by flexibility, efficiency, reliability and scalability, which is quite suitable for constructing information recommender system.In this thesis, the author firstly presents the framework of an intelligent information recommender system based on Agent. The model has three levels, each of which has a kind of agent that undertakes the tasks of interface providing, information filtering and interest learning or information retrieval.Secondly, the author expatiates on the information filtering strategies and interest learning strategies. Hybrid filtering strategy is proposed by combining content-based filtering and collaborative filtering. Content-based filtering employs keywords vector space model; collaborative filtering employs and expands the evaluations matrix by...