Binary Information Press
Traditional meta-search engines are not well considered user's interests, unable to provide users with personalized service. This paper presents a user personalization mechanism in agent-based meta-search engine. The mechanism can create and update user pro le by mining user behavior during information retrieval, it can well reflect user's current interests and be able to update the user's interests dynamically. Use user pro le in filtering the search results, can make the search results meet user's requirements personalized.