Characterization and Modeling of Time-Varying Networks

Abstract

The interconnected nature of our daily lives, both virtually and physically, highlights the importance of understanding temporal networks in the context of epidemic and information spread. This dissertation aims to address this challenge by proposing characterization methods for temporal networks. In Chapter 2, the analysis reveals that close temporal contacts are generally close in topology, with virtual contacts showing a stronger correlation, suggesting the potential for social contagion. However, a limitation is acknowledged, as the methodologies assume interactions only occur between pairs of nodes.

Chapter 3 extends the focus to characterize temporal higher-order networks involving groups of nodes larger than pairs. Findings demonstrate differences between collaboration and physical interaction networks, with physical contacts exhibiting strong correlation between topological distance and temporal delay. In contrast, collaboration networks show weak or absent correlation.

Considering temporal networks as spreading processes, Chapter 4 introduces a methodology to identify underlying spreading processes among nodes, specifically exploring the congestion contagion of airports in the U.S. air transportation network. The proposed heterogeneous Susceptible-Infected-Susceptible (SIS) spreading process effectively reproduces nodal vulnerability and outperforms a homogeneous model.

The dissertation concludes with reflections on the insights gained and suggests future research directions in the field of temporal network characterization.