Web-based User Preference Mining And System Implementation

The amount of Internet data and users have explosively grown with the rapiddevelopment of Internet, users often get lost in the web because they should spend moretime to find the information. Then people lost contact with the valuable web. So how toprovide users personalized service based on web and users’ interest has recently become ahot research topic and wide application.In this thesis we focus on user preference model based on web, and also we work outa platform to provide personal service. Firstly we introduce the background of the researchand the existing domestic and international related work on this issue. According toanalyze the existing individual service system and user preference model, we realize thatthe existing model of user preferences is not good at dynamic tracking users’ interest.To reflect the dynamic preferences of users’ interest accurately, we propose a userpreference model, which divide users’ interest into long-term interest and short-terminterest. We also introduce the study of interest in psychology and propose the concept offreshness in our model. Our model is not only concerned about the state of the users’current interest, but also concerned about the long-term interests, and pays more attentionsto the dynamic changes of users’ interest. We also propose recommended model to providecustomer personalized service based on our research about users’ interest.Finally, we work out a platform to provide personal service based on our model. Wedesign and implement our system based on data of the largest web community platform inour country. The experimental analysis verifies the accuracy of our model which caneffectively reflect the dynamic preferences of users’ interest.