Design And Implementation Of Users' Behavior Analysis System On The Reading Business Of Mobile Internet

With the rapid development of mobile Internet technology and new services, the network data grows explosively which brings shocked impact on traditional business of the major operators and put enormous pressure of management work on the Internet Service Provider. In this case, the major operators put forward a new requirement called intelligent pipeline which is badly in need of analysis group composition and preferences of user's and provide users with a more personalized service. It is significant for operators and Internet Service Provider to get information of the Mobile Internet user behavior characteristic. This thesis focus on the accurate identification and monitoring of mobile Internet user behavior characteristics and achieve a based on deep packet inspection technology of mobile Internet user behavior analysis system, based on the traditional monitoring system and took the mobile reading as the breakthrough point.The main contents of this thesis are as follows: Firstly, a set of design ideas and implementation methods of user behavior analysis system are proposed for mobile reading business users,which has a certain reference value by studying the status of domestic and international user behavior analysis for internet service providers and telecom operators; Secondly, realizing distributed topic crawler based on Hadoop, and make use of distributed topic crawler to access novel website of the internet for obtaining detailed information of Cyber-fiction by studying Hadoop distributed system framework and distributed topic crawler; Thirdly, adopting the DPI technology to identify the Mobile Internet business, and divided into real-time DPI service identification and Senior DPI service identification for meeting the needs of different business identification. The real-time DPI service identification primarily responsible for identifying the type of business and traffic statistics, and Senior DPI service identification primarily responsible for identifying business details, such as name of books, type of books, author and other information; Fourthly, corresponding index data is count and analyzed according to the different thematic applications, and puts forward a mobile reading user preference program analysis algorithm combining standard deviation and H-Index algorithm, which can mine reading habits and preferences of mobile Internet users from the mass of user data.This thesis use the data as experimental data,which was collected from the Mobile Internet, then verify the function and performance of the system,and demonstrate the application analysis of several special topics. Services could be identified in this system through vast amounts of user data, and user's reading behavior characteristics is summarized with user preferences algorithm, and it has reference value for each internet service providers and telecom operators to develop the corresponding marketing strategy.