The Research And Design Of Personalized Internet Information Retrieval System

The dissertation suggests a Personalized Internet Information Retrieval System (PURS), which integrates Internet information inquiry and collection so as to establish a new type of user interest orientated information service system. The PURS can attain the automatic identification of user interests, the automatic formation of user model and meanwhile the system can assist users in the construction of retrieval request, the inquiry of information and the acceptance of pushed information. Moreover, the system can measure the match between returned results and user interests based on the user model, and achieve full text supply function.The dissertation is divided into three parts totally, comprised of eleven chapters. The first part (Chapter one and Chapter two) is an introduction, in which the author discusses the present situation of the Internet exploitation and utilization and then analyses existent problems therein (Chapter one). According to what is mentioned above, the author discusses the procedure of personalized Internet information retrieval and proposes design thought and principle of the PIIRS, makes the overall structure of PURS, and analyses for feasibility. The second part (Chapter three to Chapter six) is study of key technologies, including study and analysis of user modeling, machine learning, search engine, intelligent agent, web page recognition, information filtering, data mining, man-machine interaction etc.. The third part is implementation. The author implements the five subsystems of the PIIRS, which are User demand and interest description subsystem, information acquisition subsystem, information presentation and feedback subsystem, subject mining subsystem, and management and schedule subsystem.During the procedure of system design and implementation, the author has made some innovative efforts such as: (D establishing the user interest orientated model, the model receiving user interests continuously and conjecturing user interests by interaction with the user, accumulating user preferences in information demand, thereby achieving self-adaptive retrieval, ﹑roviding a feedback method which is based on the human-machine interaction, summarizing the user operations on the interface of result presentation, and designing an algorithm for capturing user operation behaviors, by which the changes in user interests and preferences can be learned potentially, ﹐ffering a method for userinterest mining which can extract subjects of information confirmed by user, thereby conjecturing or predicting different kinds of expressions of the same interest or extracting the new interests or unexpressed interests, ﹑roposing a solution of personalized Internet information retrieval based on the user interests in accordance with the above-mentioned work, the solution having very strong feasibility and practicality with taking user interest model as center, employing machine learning (active learning and passive learning) and data mining as tools, and being assisted with network robot,.The dissertation concludes the features of PIIRS, and points out insufficient places.