Agent Based Individualized Internet Information Retrieval System

In order to satisfy user's demands for high-quality Internet's information, Personalized Internet Information Search has become one of the hottest research subjects.Based on the research of search engine's principle and method, this thesis proposes an Agent Based Individualized Internet Information Search Model, which combines the Full Text Search Engine, Search Index/Directory and Meta Search Engine. The model establishes and maintains User Profile through Tree Index/Directory and Agent and can finish the task of individualized information retrieval well.Do the innovation and improving to the key technology that the Information Search Agent involves: Weighted VSM (Vector Space Model) Correlativity Algorithm can measure the correlativity between the Web page and the topic more accurately; Subject Based Robot Crawling Algorithm can make the Robot crawl more meaningful; the new realization of Document Selection Algorithm improves the performance of another Web information search method-Meta Search, making it capable of finding the most useful material from numerous search results for user.Provide an Agent Based Personalized Internet Information Retrieval System realization scheme and test the realized system. Test data show the efficiency of recommendation, hit rate, systematic expansibility and interface friendship are satisfactory. Due to the complexity of the subject, the Document Selection Algorithm of Meta Search and User Profile need more efficient and accurate design and realization. The Subject Based Robot Crawling Algorithm needs to be proved that it can reach the most correlative Web page and the process of the crawling converges rapidly with the increase of the hyper links' depth.