Research On The Presentation And Updation Of User Profile For Personalized Services

Rapid growth of Internet information lead to rich information but lack of knowledgerelatively. How to seek the necessary knowledge from the information ocean accurately andquickly restrict the efficient use of network increasingly. In order to adapt to different userneeds, personalized information services can provide users who has different personalitywith targeted services and strategy. As the key technology of personalized informationservices, the quality of user’s interest modeling put a direct effect on the performance ofthem.The user’s interest model is the basic and core component in a personalized servicessystem, but user’s interest will change over time, interest drifting problem cause thedecrease of the prediction performance. therefor, this paper presents a user interest modeland the approach for dealing with the interest drifting problem based on a hierarchicalvector space model, term form interest topic, user profile is formed with interest, thenhierarchical VSM set up based on the user profile. To estimate user interest of Web pagesquickly, Adaptively update interest model based on the user’s feedback. Experiment resultsshow that the design of the user model and the update mechanism that can effectivelyimprove the prediction performance for personalized services, with improvement onaccuracy and recall.The main works of this dissertation are as follows:(1) The form of personalized services is described firstly. To be the core ofpersonalized service, user modeling has recently attracted much attention. The techniquesof user modeling are induced, mainly aims to investigate to how to organize the user’sinterest sources effectively and how to use user’s feedback, user profile representations,document computing, interest updation, as well as methods of evaluation.(2) A method that constructing user profile based on hierarchical VSM is proposedbased on analyzing traditional user profile representations. term form interest topic, userprofile is formed with interest, then hierarchical VSM set up based on the user profile.Compared with traditional integral representation, this hierarchy show better performance.(3) A adaptive updation mechanism for user profile is designed based on analyzing theclassical algorithms about interest drifting for update user interest dynamicly. Finally, thecomparative experiment on user’s interest modeling and updation are performed. We compared the performance based on different methods of feature selection and updationalgorithm, also analyzed experiment results. Which show that TFIDF is superior to TFmethod for improving system performance, adaptive updation algorithm for user profilebased on user feedback enhanced expressive ability of user profile effectively. It is provedthat the user modeling and the renewal realization in this article are realizable and reliable,it has the practical application value in personalization system.