Analysis On The Behavior Of E-Business Users Based On Telecom DPI Data

With the rapid development of the Internet and the continuous upgrading of Web technology, people's daily way of life more and more shelved into the Web. With the coming, is the advent of large data age.Big data comes from a wide variety of behaviors that are generated when a large group of Web users conduct daily network access. Rich information and value is behind the user behavior, which has had a huge impact on the enterprises operating mode. Network operators through the DPI technology collect and analyse network traffic and mine the behavior of E-Business user, but the huge flow of traditional methods and models put forward a difficult challenge. Utilizing parallel computing and distributed storage technology, the Hadoop platform provides an alternative to massive DPI data processing. The main purpose of this paper is to use DPI data as the base data, analyze the behavor of users accessing E-Business through Wireshark tool, extract the feature vector of E-Business user behavior, and construct the user profile based on Hadoop platform by MapReduce programming.The main contents of this paper are as follows: (1) The overall design scheme of the behavior analysis system for large-scale E-Business users is given. The system is mainly composed of basic data acquisition module, DPI data preprocessing module and user behavior analysis module. And the realization of these three modules is given. (2) The user behavior feature vector extraction method based on MapReduce programming is realized according to user geographical characteristics,user terminal-agent characteristics, access to different E-Business websites ; (3) Through DPI and E-Business product-information data aggregation, it realizes the construction of multi-dimensional dimensions of electronic business user profiles in Hadoop platform; (4) According to the user's profiles, a user interest recommendation model based on ItemBased collaborative filtering is proposed.