Towards the next generation social network

Online Social Networks (OSNs) is part of everyday life for many people, but many question them because they fail to preserve the users' privacy. Therefore scientists tried to propose distributed architectures for the implementation of OSNs, giving birth to Distributed Online Social Networks (DOSN). To fully embrace the decentralization, the knowledge of how people use OSNs is needed, and in this thesis, we propose analyses to cover the lack of knowledge and contributions towards a next-generation DOSN. We start by analyzing how community detection can be beneficial to OSNs in a static and dynamic fashion, and design a privacy policy recommendation system. We then propose Incremental Communication Patterns to capture malicious users, such as bots or stalkers. We also turn our attention to the scenario Online Social Groups, in which we study the interaction structures of its users. To support the decentralization, we propose an innovative social overlay, called Contextual Ego Network based on contexts, a distributed dynamic community detection and management protocol, and a study of the InterPlanetary File System, and discuss their application in DOSNs. Lastly, we focused on Blockchain Online Social Media by taking Steemit as a case study. We started by studying the interaction graph and the follower-following graph of the users. Additionally, we analyzed the features of the users, gaining insights concerning the topics discussed and the behavior of block producers and bots.