Identifying website communities in mobile internet based on affinity measurement

With the rapid development of mobile devices and wireless technologies, mobile internet websites play an essential role for delivering networked services in our daily life. Thus, identifying website communities in mobile internet is of theoretical and practical significance in optimizing network resource and improving user experience. Existing solutions are, however, limited to retrieve website communities based on hyperlink structure and content similarities. The relationships between user behaviors and community structures are far from being understood. In this paper, we develop a three-step algorithm to extract communities by affinity measurement derived from user accessing information. Through experimental evaluation with massive detailed HTTP traffic records captured from a cellular core network by high performance monitoring devices, we show that our affinity measurement based method is effective in identifying hidden website communities in mobile internet, which have evaded previous link-based and content-based approaches.