Algorithms and Dynamical Models for Communities and Reputation in Social Networks

A persistent problem when finding communities in large complex networks is the so-called resolution limit. This thesis addresses this issue meticulously, and introduces the important notion of resolution-limit-free. Remarkably, only few methods possess this desirable property, and this thesis puts forward one such method. Moreover, it discusses how to asses whether communities can occur by chance or not. One aspect that is often ignored in this field is treated here: links can also be negative, as in war or conflict. Besides how to incorporate this in community detection, it also examines the dynamics of such negative links, inspired by a sociological theory known as social balance. This has intriguing connections to the evolution of cooperation, suggesting that for cooperation to emerge, groups often split in two opposing factions. In addition to these theoretical contributions, the thesis also contains an empirical analysis of the effect of trading communities on international conflict, and how communities form in a citation network with positive and negative links.

Autorentext
After having obtained a (cum laude) Masters in sociology from the University of Amsterdam, Vincent Traag turned to the Université catholique de Louvain in Belgium, where he obtained his Ph.D. in applied mathematics, under the supervision of Paul Van Dooren and Yurii Nesterov. Trying to combine sociology and mathematics, Traag performed extensive research in complex networks, with a particular focus on communities and reputation in networks. One particular interest is how conflict and cooperation arise. In his current position Traag is trying to understand how elite networks in Indonesia reflect political events.

Inhalt
Introduction.- Part I Communities in Networks.- Community Detection.- Scale Invariant Community Detection.- Finding Significant Resolutions.- Modularity with Negative Links.- Applications.- Part II Social Balance & Reputation.- Social Balance.- Models of Social Balance.- Evolution of Cooperation.- Ranking Nodes Using Reputation.