Abstract
Social interest refers to a kind of preference that an individual enjoys in social networks. In real world, people’s social interests are influenced by various factors such as social relationships, historic social interests and users’ private attributes. However, few publications systematically study the problem of social interest inferring in a real mobile network. In this work, we study the extent to which one’s social interest can be inferred combining these above factors together, and investigate the Time-Space Probability Factor Graph Model (shortly, TS-FGM). We propose a novel social interest inferring approach based on TS-FGM and transfer the problem into the maximizing problem of the objective function of TS-FGM. In our approach, historic social interests are formalized as time factor, social relationships are formalized as space factor, as well as users’ private attributes are formalized as attributes of network nodes. We validate the presented model on an real mobile communication network and find that it is possible to approximately infer 81% of users’ social interests.KeywordsSocial Interest InferringFactor Graph ModelMobile Networks
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a call