Research Of Information Push System Based On Agent

In nowadays situation of network resource growing rapidly, to enable users on the Internet not only spend less time and energy, and to make timely and effective access to useful information, this paper introduces a model of the information push system based on multi-Agent that use information push technology, machine learning and intelligent Agent technology. The system can change the traditional manner of capturing information by users. It provides information to the users by learning and adjusting their interest model.The framework of information push system based on Agent has three levels that contain information retrieval, information filtering and information denotation, each of which has kinds of agent service for users, such as information retrieval, information filtering, interest learning, provide interactive interface and so on. Then the paper expatiates on the interest learning strategies and information filtering algorithm, and sets up user interest model with Vector Space Model, searching information by using word segmenting and searching engine, filtering information using a new hybrid filter algorithm. Based on similar item-set of target item by combing user-based collaborative filtering, the algorithm solved the problem of the lower quality of recommendation information brought by sparsity with content-based filtering.Finally, the author designs a system by using Java language on the Aglet platform. The system achieves the desired goal basically after continuous testing. At the last part of paper are the conclusion and future works.