Stochastic dynamics on networks: insights into opinion polarization

Stochastic dynamics play an important role in modeling various real phenomena, including collective behavior in social networks. We explore mathematical methods for studying these dynamics, focusing on two main approaches: individual- and compartmental-based. While the former tracks each node's state, the latter aggregates nodes based on structural properties like network connectivity. We assess the accuracy of these methods in different network conditions, highlighting challenges ahead.

As a case study, we examine a follower-followed network extracted from Twitter, analyzing the community structure of French users, including members of Parliament and regular users. By employing network embedding techniques and political survey data, we map the network onto a four-dimensional opinion space, revealing cohesive groups with similar political attitudes. Our study shows that politically extreme communities segregate from other groups, indicating the level of polarization present.

Presential in the seminar room. Zoom stream: 

https://us06web.zoom.us/j/98286706234?pwd=bm1JUFVYcTJkaVl1VU55L0FiWDRIUT09