Design And Implementation Of User Interest Extraction System Based On News In Mobile Network

Personalized recommendation technology can automatically search information, which meets users’demand in the network for different users with different preferences, to save manual search time for users. How to extract the user’s interest characteristics and establish user interest-model is the core and foundation of personalized recommendation technology, which directly has impact on the quality of recommendation. In different network environment, users have different browsing habits and express their interest in different ways, so it needs different methods to extract user’s interest. The recommendation systems based on mobile internet have their own characteristics, which differ from the recommendation systems based on traditional network, thus, an appropriate model according to user’s feature in different network can improve the accuracy of user’s interest expression.By the analysis of the mobile network’s characteristics that are different from the characteristics of traditional network, this dissertation researches user interest modeling mechanism, then designs and implements the user interest extraction system based on news in mobile network. The research points and contributions are listed as follows:(1) It investigates and analyzes the applications of personalized recommendation technology at home and abroad, focusing on analysis of user interest modeling theory. User interest modeling consists of the collection of user interest data and user interest-model representation which are summarized in this paper.(2) It analyzes the characteristics of mobile network environment and the features of news that is deliveried through mobile network. In mobile network, location is a key element to find user’s feature and mobile users can access network at anytime and anywhere. Moreover, there are a variety of mobile terminal equipments that are different from each other, and the scenarios of mobile users changes rapidly. These feature of mobile network can help dig out user’s interest. News is up-to-date and updates faster. So, user model representation and personalized recommendation algorithm for personalized news service should be designed according to the characteristics of news.(3) With the characteristics of mobile network environment and news, it proposes the user interest-model representation and updating mechanism that integrates of scenarios information. User interest-model is represented with vector space model, and it measures the degree of user’s interest in news page through user’s browsing behavior. Finally, the time decay mechanism is applied in the process of updating the user interest-model to highlight the transfer direction of user’s interest.(4) it designs and implements the user interest extraction system. It is divided into interest data acquisition module, interest model update module, interest analysis module and management module. Interest data acquisition module locates on the client, whose function is to record the user’s scene and browsing behaviors in time, and then, transmit the data to the server-side asynchronously. In interest model update module, user’s browsing data is analyzed and calculated, and it mines user preferences and update user interest-model incrementally. Interest analysis module is responsible for analyzing the user interest model to obtain user’s interest structure. Management module can manage and maintain user information and system parameters configuration information. In addition, experiments are done to evaluate the system, including both subjective and objective evaluation of the system. The experiments show that it can dig out user’s interest characteristics in different scenarios according to the user’s browsing behaviors and scene information, moreover, the user’s interest from system analysis and user’s real interest are in good conformity degree.