Community evaluation in Facebook groups

One of the main points of the Next Generation Internet is to have a user-centric approach where daily behavior and social life of the users are studied and analyzed in order to model networks and services. Indeed, social life represents the general overview of the behaviour of people, because it can provide information about hobby, relationships, but also similarity, etc. Today, the main channels to study the behaviour of people are Social Media. A great trend in current Social Media platforms is to offer the opportunity to establish and join groups of people, which represents one of the main characteristics of offline social network, where people are clustering, usually based on their interest (work, family, etc.). Despite human behaviour in current Online Social Media have been studied in depth, characteristics of online content-based social groups are still unknown. In this paper, we investigate whether communities can be recognized also in groups defined by users of Social Media platforms and we study how these communities evolve over time. For this purpose, we exploited a real Facebook dataset which consists of 18 Facebook groups of different categories and 3 different community detection algorithms. Our results provide important insights about the behaviour of users in the context of social groups and reveal that the majority of the groups present interactions-based communities, and in particular there is one massive core community which attracts other users and communities.