“From social network data to social mechanisms: a data-driven agent-based modeling approach”

Abstract: Social relationships can be represented as signed networks, in which positive and negative links represent friendship and conflict. But what can the observed structure of these networks tell us about the mechanisms shaping social relationships?

In this talk, I will describe the development and application of an agent-based model combining two influential theories of signed social relations: structural balance and status theory. After exploring the model’s behavior on synthetic networks, I will turn to empirical data and the challenge of estimating model parameters from observed network patterns. I will discuss our approach to model calibration, the difficulties of comparing parameters across different datasets, and the advantages of working with a collection of social networks from Spanish high schools.

Beyond providing networks for calibration, the school dataset contains additional information about relationship strengths. This allowed us to investigate associations between the inferred importance of status and weak ties and motivated further research suggesting a possible pathway from weak ties to structural polarization. This talk will illustrate how empirical data can both constrain theoretical models and open up unexpected scientific questions.

Speaker Bio: Piotr Górski is an Assistant Professor at the Faculty of Physics, Warsaw University of Technology, Poland, and currently a Visiting Scholar at the University of Washington’s eScience Institute. His research focuses on complex networks and computational social science, particularly the mechanisms shaping social relationships and collective behavior. He combines agent-based modeling, network analysis, and empirical data to investigate structural balance, social status, and polarization. His research has been published in journals including PNAS and Physical Review Letters.