Research And Design Of User Behavior Analysis And Classification In Universal Network

The current open Internet architecture has promoted the vigorous development of the network and brought a number of problems as well, such as the poor security mechanism etc. Therefore, the Universal Network was proposed creatively which separated the whole network into two-layer architecture. By the mapping mechanism between AID and RID, the Universal Network separated the user identity information and location information and ensured the security the network. However, as a new kind of network architecture, the Universal Network still needs to be improved, especially in terms of user behavior analysis and classification. Various and personal services can’t be provided to the users.To solve the above problems, a method and system of user behavior analysis and user classification is designed in this paper, based on the mature user behavioranalysis technologies, large data processing platform of the existing network and the unique characteristics of the Universal Network. By means of analyzing the construction and preference of users, mastering the rules of user behaviors, the allocation of resources can be optimized and fine-grained management can be achieved.The main research contents and results are listed as follows:(1) User behavior analysis:this part is divided into data collecting, traffic analysis and content analysis. To be specific, high-speed packets are captured by PF_RING. Then NetFlow reads the offline packets and restores to the session level. IPFIX analyzes the protocol by plug-ins. At the same time, the distributed platforms based on MapReduce achieve the analysis of the content of user behaviors.(2)User classification:put forward a standard of user classification, design a user classification framework, build the user modeling based on the basic attributes and behavior attributes, and take a deep digging in user behaviors by the clustering algorithm with MapReduce. Finally, the prototype system platform verifies the above content. What’s more, the results of user behavior analysis and user classification are shown by the Web interface, including the user traffic statistics, user distribution, website ranking, and the hot topics and so on.