Interactions around social networks matter: Predicting the social network from associated interaction networks

Abstract

Tie formation in social networks is driven by different motives that are not always apparent in the social network itself. These motives differ from one social network to another, depending on, e.g., the network's purpose, such as advice seeking or collaboration, and the effort it costs to establish a friendship relationship. A common factor that exists in almost all social networks is homophily: the tendency of social network members to connect to similar members. In this work, we look at the tie formation process in social networks from a different perspective where we consider not only a social network SN, but also a set of associated interaction networks Gn around it. We show, based on 6 social networks and in total 20 different associated interaction networks, that it is possible to predict the entire social network's structure to a satisfactory extent, only by knowing the structure of these interaction networks. As social networks are based on a voluntary relationship while some of the interaction relationships are at most semi-controllable for most members, e.g., being together in a team, this seems to indicate that whom we choose as a friend is also determined by whom we interact with.