Personalized user profile mining based on user's browsing behaviors

User profile that reflects the user's interests forms the basis of providing personalized information services, which plays an important role in the personalized information system. User's daily browsing behaviors can reflect the user's interests to some extent. User interest taxonomy tree is introduced to signify user's interests in this paper. A plug-in embedded into Internet Explorer is developed for capturing user's browsing behaviors. These browsing behaviors including browsing time, and the number of underlines, annotations, highlights, bolds and italics that are marked up in each browsed web document, are recorded and formalized by the behavior table. Thus, user's interests can be mined from these browsing behaviors. To make user profile can accurately reflect user's current interests, an adaptive user profile based on memory model which can be updated with time automatically is presented in this paper. An experiment is designed to verify our proposed approach. Experimental result indicates that user profile based on our approach can reflect user's interests accurately.