POSTPONED DUE TO ILLNESS. Public defence in Computer Science, M.Sc. (Tech) Javier Ureña Carrion

Title of the thesis: Structured Interactions: inferring social behavior in networked systems

Doctoral student: Javier Ureña Carrion
Opponent: Professor Esteban Moro, Network Science Institute, Northeastern University, USA
Custos: Professor Mikko Kivelä, Aalto University School of Science, Department of Computer Science

Interactions are the basis of our lives. Some interactions make up relationships that span lifetimes, while others are as fleeting as liking a stranger's post on social media. However, interactions don't exist in a vacuum, but reflect social structures. This thesis offers insights into how human behavior both shapes and is shaped by social structures, drawing on methods from network science and theoretical frameworks from sociology. Such insights stem from the analysis of large-scale communication data and mathematical models of behaviour, two key areas of social network analysis.

The first area focuses on the footprints left by phone calls or emails. These datasets have been vital for understanding social systems, from contact patterns to information spreading. We explore how social structures are reflected in communication systems. For example, one key result shows how "social times" affect the composition of communication networks. Relationships where people only talk during working hours have different structural roles than those where contacts extend into evenings and weekends. The former is more likely to act as a "bridge" in the network, connecting individuals that don't share common friends, while the latter tends to be surrounded by friends.

The second area examines how segregated networks or group dominance can arise from the simple ways people make new connections. We propose a mathematical model to show that meeting friends of friends can be a venue for local homogenization. This is because the friends of your friends tend to be more similar to you, leading to more in-group connections which can turn into network segregation.

In large systems, people with many friends can more easily reinforce their popularity. This helps explain how social media influencers attract more followers. We show that this dynamic can also affect groups, leading one group to act as a core, while the other as a sparse periphery. While in our model such dominance exists even when each group has a strong identity, we find that in many real-world cases, such as politicians in social media, only the core favors it's own group. As a result, small, cohesive groups can dominate discussion systems and lead to centralized networks even if, in theory, anyone can interact with anyone.

This thesis shows how networks of human interaction are deeply intertwined with social structures, with potential applications in social media design, epidemiology or public policy.

Thesis available for public display 10 days prior to the defence at: https://aaltodoc.aalto.fi/doc_public/eonly/riiputus/ 

Contact information:

Doctoral theses of the School of Science: https://aaltodoc.aalto.fi/handle/123456789/52 

Zoom Quick Guide: https://www.aalto.fi/en/services/zoom-quick-guide