Empirical analysis of synthetic and real networks

Abstract

With increasing digitization a wide variety of systems from diverse domains such as computer science, transportation, social science have become available in the form of networks. It is argued that to understand complex systems a deep understanding of the networks behind them is needed. A network theoretic perspective provides valuable insights into the structure and trends of systems. Data-sets belonging to different domains have their own unique features and behavioural trends and the current inquiry aims to highlight this. In this inquiry, a comprehensive analysis of synthetic and real-world published benchmark data-sets, evaluation methods, and open source projects is performed. The aim is to provide novice and expert users with tools for algorithmic designs and methodologies. Empirical studies are used to compare the performance of network theoretic tools on common data-sets. Finally, limitations of the network perspective on systems are listed and research directions to facilitate future study are elaborated.