Multi Dimension Analysis Of Network User Behavior

With the rapid development of Internet, Internet services are becoming more and more active than before, such as personalized service. To develop personalized services, it is necessary for companies to have a deeper understanding of user’s interests. User behavior analysis can find out potential features from large amount of raw data and provide decision support for business and scientific research.In this paper, we analyze the behavior characteristics of users in multiple dimensions and mining potential user groups in each dimension as well as the corresponding habits and regulars. Through the analysis results of several dimensions, we analyze the comprehensive behavior patterns and characteristics. The data used in this paper come from the real Internet data of Beijing Institute of Technology. The original data cover a large user community, so the analysis results can reflect the characteristics and behavior of the broadband Internet users in the university campus. The analysis model and method of this paper can also be applied to the analysis of user behavior in other metropolitan area network.Specific research contents of this paper include:(1) The analysis of internet active behavior dimension. In this dimension, we analyze the behavior characteristics and patterns of Internet users. Bcv-k-means clustering algorithm is used to dig out the behavior patterns of users in the active dimension.(2) The analysis of web access interest dimension. In the analysis of this dimension, we proposed an automatic clustering algorithm based on entropy for discovering the interest pattern over users’ web log. We introduced the information entropy on the basis of clustering algorithm. Compared with traditional clustering algorithms, our method does not require any parameters specified by the end user. Meanwhile, it can discover the clusters in arbitrary shape and size. The validity of the algorithm is verified by experiments.(3) The analysis of network application behavior dimension. In this dimension,we analyze the commonly used network application of the user, then mining the user’s network application patterns and characteristics of network applications use.(4) Correlation analysis of dimensions. In this paper, we discussed the correlation of analysis results, and we carry out supplementary analysis to the special user patterns.