A Research Of User Interests Based On Browsed Content

The rapid development of information technology and the Internet had provided a lot of convenience to people’s life, but the web users spent too much time to get their own requirements from the vast data of Marine information. Personalized service is came into being in this context. The research of users’interests is a necessary component to provide personalizing service, and the accuracy of user interest model directly determines the quality of personalized service. So it has certain theory significance and practice value to analyze and study the users’interests based on browsed content. The research content includes:(1) The related theory and technology have been studied. Studying lots of related work, research the method of collecting the user interest information, the representation methods of user interest model and the key technology to analyze the content of user interest.(2) Proposing a hybrid measurement method to discovery user interests. To solve the feature dimension disaster problem, this paper proposes an algorithm of weighted summation which combined information gain and mutual information to reduce feature dimension. Because of the difficult to get accurate users’interests, this paper puts forward a new method which classifying the browsed pages before clustering to get users’interests. Finally according the retention time of user page and the length of content to calculate the degree of interest, and in combination with characteristics to present the user interest.(3) Completing the design and implementation of user interest system. Describing the system’s overall design framework, and dividing into four modules:user interest content collection, data pretreatment, the user interest discovery, user interest representation. The function of each module and design are described in detail, and demonstrate the realization of each module.