Observing Friendship Patterns Through Smart Phone Radios

Nowadays, social networks are not used only to analyze the human society but also to rebuild it. Social activities and interactions increase dramatically. A common way to construct the networks is to generalize a single dataset, such as location or proximity data. It, however, is very hard to find a universal method obtaining highly accurate networks which represent the social relationships. The structures of the networks may also vary depending on definitions of nodes and edges. Previously, the relations between human mobility and human relationships were mainly studied so far. In this paper, w e design a combined network model with multiple datasets to provide a highly efficient way for constructing social networks. Specifically, the networks are friendship networks and individual activity networks. The effect of friendship on social activities and interactions is analyzed as well. The performance of the model is evaluated using centralities and coefficients. Finally, relationships among the networks are also shown.