Regensburg 2025 – wissenschaftliches Programm

Bereiche | Tage | Auswahl | Suche | Aktualisierungen | Downloads | Hilfe

SOE: Fachverband Physik sozio-ökonomischer Systeme

SOE 3: Poster

SOE 3.5: Poster

Montag, 17. März 2025, 17:30–19:30, P4

Interplay between algorithms, cognition and social interactions — •Beniamin Sereda and Janusz Hołyst — Faculty of Physics, Warsaw University of Technology, Koszykowa 75, PL-00-662 Warsaw, Poland

Modern societies are increasingly facing the issue of opinion polarisation, amplified by recommendation systems and their role in forming information cocoons. To understand how the evolution of user preferences and the topology of social networks influence opinion dynamics, the integrative agent-based model DAISY was developed. This model assumes three distinct levels of interactions: algorithmic, social, and cognitive. Algorithmic interactions are governed by recommendation algorithms, the social level reflects user preferences regarding whom to contact in the context of recommended items, and the cognitive level describes how individuals process information.The dynamics of preferences are driven by the algorithmic level, while opinion dynamics occur on the social and cognitive levels. Co-evolutional perspective on dynamics allows observation of how changes in network topology, triggered by recommendation systems, shape opinion evolution. Simulations show that increased personalisation accelerates the formation of informational cocoons, though in a non-linear and complex manner. Analysing social network evolution also revealed a deeper understanding of the irreversibility of processes within the system.

Keywords: information cocoons; opinion polarisation; recommender systems; opinion dynamics