Online Social Network with Communities: Evolvement of Within- and Cross-Community Ties

Abstract

More and more online platforms allow users to not only befriend each other but also form groups or communities. While communities can encourage user engagement, they also bring up the concern of fragmenting a social network, weakening its value added for the platform. In this paper, we analyze a unique dataset from an online game, which includes multiple snapshots of the players' social network and their community memberships. We propose and estimate a model to capture the evolution of social network with communities. The model allows for mutual influence between friendship and community choices, as well as rich unobserved heterogeneity at the individual level. We find that ties are much more likely to form within-community than cross-community. However, in the longer term a great fraction of cross-community ties actually come from within-community tie formation, thanks to the mechanism where individuals maintain ties with previous communities when switching communities. A key factor in this mechanism is mobility, the frequency that individuals switch communities. By using the appropriate policy to balance the overall mobility, the platform can harness tie formation within communities to achieve stronger connections across communities.

Keywords: social networks, network dynamics, community structure, online communities, within- and cross-community ties, network fragmentation, individual mobility