Graph-Based Pattern Mining And Application

Nowadays, the World Wide Web (WWW) system is developing rapidly both in the depth and the extent; web has become a huge source of the information. Our life is changed unprecedented by the Internet, and WWW browsing has become an important part of our daily life. But there are still two familiar problems in the WWW browsing: one is it's hard to search and acquire the useful information and the other is the information foraging behavior is inefficient. The traditional website (while its structure is static) can't meet people's requirements: people want to find their useful information efficiently and accurately, and they want to enjoy personalized service. Personalized website research can improve the overflow of web information and the lack of personalized service. How to set up a personalized website to adapt to all kinds of user needs have become a jumped-up important research area in the world. This research has important theoretical significance and practical value. In this thesis, based on the Web usage information, Data Mining techniques are used to research on personalized service problems, mainly in the following directions:First of all, as the beginning and basis of research on Web Usage Mining and personalized service, this paper first carried out a comprehensive analysis and discussion in various stages of Web Usage Mining as well as Data Mining techniques, studied related theory of Web Usage Mining and personalized website, prospected the research directions of Web Usage Mining in the future.In the next place, in need of Web Mining and site personalization, this paper used weighted directed graph to indicate the site structure, estimated the probability of mutual cited between each page through the user's access records and achieved user clustering at the same time, ultimately calculated the authority pages of the relative groups of users through improved PageRank algorithm, provided the basis for achieving site personalization.In the end, in the configuration of the framework of personalized site system, designed a integrated framework of personalized site system based on WEB. The system is a continuous cycle process, firstly collected data through the web site, then carried out data mining with the data of site framework and the data which after the necessary data pre-processing work, put the acquired knowledge into application finally. The entire framework can be applied to existing WEB site, so that the site would have personalized features.