The Study Of User Interest Model In Personalized E-commerce System

With the development of the Internet technology, e-shopping becomes more and more popular, and personalized recommendation systems are also become increasingly important. However, the traditional recommendation strategies based on users' collaborations can not generate an explicit user model. Recommendation is based on direct comparisons of the user similarity, and the recommending result is not so valuable. Other systems based on ware item recommendation can overcome shortcomings of the traditional user collaboration recommendation, but at the same time lose recommendation of the user's interests.In this paper, with the background of online bookstore application, by investigating user interests, a new user interest model is established based on fuzzy set theory, and a new book recommendation algorithm is proposed based on this model. A user interest file is established after the user's login, then the user's interest is learnt through the user's shopping history and the user interest file is updated. Book recommendation is made based on the user interest file.Because of the characteristics of intelligence, autonomy and collaboration, Agent becomes a hot research area of the next generation e-commerce. In this paper, implementations the proposed model is based on the JADE platform (an Agent development platform). A prototype system is built with three types of Agent: user interface Agent, user interest processing Agent, and book recommendation Agent. User interest model and book recommending algorithm are implemented by collaborations of these three types of Agent. Finally, a simple test experiment is made for the prototype system.