On the Privacy and Utility of Social Networks

Abstract

Data that is produced on a large scale from online social network sites brings challenges to the exploration of data utility, as well as the protection of data privacy. We look into both problems from a graph perspective. In particular, we focus on community detection and graph anonymization. Graphs analyses facilitate the study of relationships or social interactions of online social networks modeled as graphs. Community detection constitutes an important tool for the analysis, by exploring the network structure and associated information. It provides insights into the network characteristics and structural properties, and thus, the social phenomena that take place. Graph anonymization approaches on the other hand protect data from private information disclosure, prevent malicious use and ease the user concerns. In this thesis, we discuss the concepts, related works and the problems of graph anonymization and community detection. We then design algorithms to solve the problems, respectively.