Mobile User Interest Modeling Research, Personalized Information Services

Along with 3G time arrival, and the rapid development of mobile networks, the mobile information services can provide each kind of advisory service for the user in real time, such as weather, news, transportation, stocks, entertainment and so on. The Mobile information service is becoming the new hot spot in the mobile value-added services market.If the mobile information services only provided for general business, it will not be able to satisfy the user to the information personalized demand. Mobile Personalized Information Service System can help the user fast highly effective to gain the useful knowledge and the most recent information from the magnanimous information, and meet each kind of user different personalization need. System provides the high-quality personalized mobile information service for the mobile user, that must know the user's interest and accurately to establish user interest model.This article has studied the user's interest modeling techniques, which are the basis and the core for personalized information services. After of the analysis of user modeling technology in foreign and domestic research status, this article put forward a new user properties mining system UIMining. Its function was through crawls takes the WAP page which had been downloaded and browsed by the mobile phone user, to excavate user's interest and establish the user long-term and the short-term interest model.This article first describes the research significance, introduces the status of personalized service and user's interest modeling technology at home and abroad. Based on 3G mobile networks, this article analyzes the data mining technology which is the basis for user's interest modeling technology. Then the article designs one kind of a new user properties mining system UIMining, and separately introduces function of UIMining's sub-module. The article describes the auto user modeling process,and mainly includes user's interests mining and automated user interest modeling.In the user's interest mining, firstly the paper introduces the climb to take the WAP page which had been downloaded and browsed by the mobile phone user. The article chooses Chinese grammar analysis system ICTCLAS to do the text word, which was developed by Chinese Academy of Sciences. Secondly, the article details of methods and processes of keywords extracted sub-module.Finally, the article uses an improvedχ2 statistic and word frequency statistics keyword extraction for the feature vector dimension reduction, and uses decision tree to achieve category.In the auto construction user interest model process, the article describes user's interest representation and user modeling techniques, and presents a new user interest modeling method, which is based on the density of user visit WAP web pages. The paper describes the user using this method to build a long-term and a short-term interest model, and update the user interest model. Finally, the experimental results show that UIMining system can describe the user's long-term and short-term interests effectively and accurately. It has the practical application value in the mobile personalization information service.