Modeling group users' interests from perspective of different network packet levels

In recent years, the users' interests have been widely used to customize the webpages and predict users' requests. During users' interests extraction process, log files on the web servers or users' navigation history data have been used a lot. However, little attention has been paid to the network packets which may actually have much hidden information about users' preferences and profiles. Is the network packet content a good indicator of users' behaviors? To answer this challenging question, a novel approach is presented to extract group users' behaviors through analysis of the network packets to highlight the relationships of intra-group interests and inter-group interests. The experiment results show that the relevance and correlation analysis is helpful to provide a comprehensive understanding of users' behaviors obscured in the network packets, contributing to the network security management and website impressiveness improvement.