Title:A Large-Scale Community Structure Analysis In Facebook

Abstract:In this work we present a large-scale community structure detection and analysis applied to the largest Online Social Network actually existing, namely Facebook. This network gathers more than 500 millions users at 2011. The structure and the characteristics of this social network have been widely investigated during the last years. Although some related work focuses on analyzing clustering phenomena on a small scale, this is the first large-scale study which considers a significant sample of the network. The process of data mining from the social network platform is here described, in order to clarify how the required information has been acquired avoiding privacy related issues. Data have been collected adopting two different techniques of graph mining, and are here discussed accordingly to the structural properties of these samples which have been already investigated in previous works. To the purpose of discovering the community structure of the considered samples, we devised and described two efficient complex network clustering algorithms, namely FNCA (Fast Network Community Algorithm) and LPA (Label Propagation Algorithm). Finally, results of a quantitative and qualitative analysis of the detected community structures are here discussed. It is shown that the dimension of community structures follows a power law distribution and the sets share an high degree of similarity, regardless the adopted detection algorithm.