Design Of User Behavior Analysis System

It becomes increasingly common for various industries to use information technology with its development. The information systems have collected a huge amount of data these years, including a lot of information about the customers. The number of the data is still increasing exponentially. How to accurately find out useful knowledge and discover the implicit mode of the data timely can help the enterprises to formulate the strategies of market, to keep the former customers, to attract new customers and to take the preemptive opportunities.This paper, based on such background, designing a user behavior analysis system situates between BSS/BOSS and increment services system, in a closed or semi-open network which, the security is assured; using data mining algorithms of Bayesian incremental learning to classify user data for obtaining the characteristics of user behavior and data model then using Bayesian incremental learning algorithm to optimize the model to orientate consumer behavior accurately. In addition, the user behavior analysis system can connect directly with operator's business systems. More specifically, the user behavior analysis system can help to find more potential users in the way of advertising.This paper first analyze the background that 3G networks operators are meeting brand-new business challenges, introducing the system diagram of user behavior analysis. The whole paper discussing from system analysis to realization of system functions focus on the following aspects in detail.â‘ Data mining technology used in the system of user behavior analysis as well as principal methods and general process.â‘¡Through analyzing characteristics of user behavior continually, categorizing users by Bayesian incremental learning algorithm based on the data obtained, with updating the classification models, to achieve the function of results feedback self-learning.â‘¢Through tree structure of industry knowledge, create a contribution degree atom model of users'lifecycle based on information obtained about user behavior and also build a classification vector model with multi-dimensional space for user by SVM algorithm.