Research And Implementation Of Intelligent Agent-Based Personalized Information Retrieval System

In existing Agent based personalization information retrieval systems, Agent's decision-making has been a difficult problem without good solution, and its expansibility is not so good either, therefore personalization information retrieval system based policy Agent is proposed in this paper.On the basis of analyzing traditional Agent technology, we design a policy-driven Agent frame (PDAF). Different from the general Agent frame, it makes the management of policy very easy, system more reliable, efficient and sensitive. Policy-based framework is independent of execution of the specific code, it is beneficial to divide policy engine from the special design, makes policy engine independent development, deployment and evolution. The new action about decision-making of management process can be plug-and-play by setting corresponding description in policy, accordingly, significantly improve the maintainability and scalability of system. Using policy makes personalized information retrieval system of management and decision-making very convenient. Uniform framework simplifies the establishment and interaction between each module. This paper focus on design and implementation of the user interest and information filtering Agent. User interest adopts representation mode based ontology concept. It takes fully into account semantic relation between interests of user, makes user's interest semantic more clear and representation more accurate. Filtering Agent makes full use of semantic expansion, such as category and synonym and so on, improves the amount of similarity algorithm based vector space model, makes judgment of amount of similarity more simple and effective. Simultaneity, by calculating amount of similarity between initial search results and user interest models, greatly enhance the accuracy of the final search results.Experiment proves that this paper propose the policy-driven Agent technology combined with the user interest model and information filtering technology to achieve the personalized information retrieval, compared with the traditional search engine, with better precision, and achieve a more satisfactory effect.