Research Of Personalized Information Retrieval System Based On Agent

With the population of Internet, more and more users search information with searching engines in recent years. It gets more and more difficult for users to find the information they need fast and accurately, though the searching engines are already well developed. With ceaseless development of the study in computer intelligence, agent technique, which possesses good flexibility and study characteristics, has been one of the promising study points in the field of information searching. It may solve the problems of getting information from Internet. This thesis first analyzes the problems in the Internet information searching in the past and at present, then gives an introduction of the relative techniques with artificial intelligence, expounds the characteristics and advantages of the system based on agent, brings forth an personalized system model of information searching, and takes the software structure for example to prove the feasibility of this model, then introduces the running mechanism in three aspects: information searching on internet, information modification and saving, and database access. In the following parts it discusses how to carry out this model in specific software and hardware environment. Then intelligent agent algorithm is discussed. First the Vector Space Model and similarity calculation of documents are introduced. On the basis of this, it analyzes the basic principle of documents clustering algorithm, brings forth a feedback basis searching model for information searching by this principle, and then gives a brief introduction of this model in three aspects: model construction, documents filtering and model reconstruction. It proves that this model is able to solve the problems existing in the information searching and embodies some intelligent characteristics. finally we provide the conclusion and future work.