Computational social network analysis from the perspective of statistical physics

Abstract

In this thesis, we study three different systems in relations with social phenomena. Two of them are empirical studies on social networks and the other one is a model of self-organized criticality. In Chapter 1, we introduce recent attention of physicists on social phenomena. In Chapter 2, we study time evolution of coauthorship networks. Coauthorship network is a kind of social network of scientific communities. The study contains the first observation and analysis of early stage of social network evolution. In Chapter 3, we study bipartite network of legislators and their followers in Twitter. It reveals a gap between the real ideological positions of legislators and perceived ones by Twitter users. It also shows political biases of Twitter users who follow legislators. The study is an example of the methods based on objective behavior of individuals that overcomes shortcomings of survey based methods. In Chapter 4, we study the Drossel-Schwabl forest fire model on scale-free networks. The model can be a simple model of social crisis such as financial crisis. It was proposed as a non-conservative model of self-organized criticality but has been shown that its scaling behavior is suspicious. However, the model running on scale-free networks shows unexpected scaling behavior.