Towards Generating Realistic Geosocial Networks

Abstract

The proliferation of location-based services and social networks have given rise to geosocial networks, which model not only the social interactions between users but also their spatial activities. Examples include traditional social networks extended with geo-annotated posts such as Twitter and Facebook, and networks such as Foursquare and Yelp that directly offer geosocial services. Despite the ubiquity of such networks in everyday life and the strong interest by the research community, a limited number of datasets are in fact publicly available. In view of this, we investigate the generation of realistic geosocial networks which find application in benchmarking and testing of analysis tasks, "what-if" scenarios and simulations. The contributions of our work are twofold. We first identify three types of synthetic geosocial networks which mimic the characteristics of real ones and second, we develop a prototype which combines graph and spatial generators, to construct such networks.