Activity-driven radicalization phenomena in public discussions

Friday, March 1, 2019

11.30 a.m.

ISI seminar room 1st floor

Dr. Fabian Baumann Humboldt University, Berlin

We present a novel model for the description of the adoption of extreme opinions in a population of socially interacting agents, which we term radicalization. Inspired by multiple reports of strong opinion polarization, measurable from social media data and our own empirical insights into pre-election debates on Twitter, we investigate on the possibilities of opinion reinforcement by peers. In a minimal scenario of a one dimensional opinion landscape, we introduce a simple mechanism which, in contrast to established (constructive) opinion formation models, might lead to an increase in the conviction of both interacting agents. Combined with further concepts of social influence theory such as homophily and/or a bounded confidence of interaction, we aim for two qualitatively different outcomes: (1) a global consensus and (2) a separation into opposed polarized subpopulations.
With respect to modern communication platforms, such as Twitter, where the spreading of information among a large number of peers became possible at low cost, we want to extend the modeling framework for a temporally changing substrate of interactions. Using the activity-driven model as a null model of temporally uncorrelated pairwise interactions we aim for an explanatory link between the observed political opinions and the (heterogeneously distributed) activities of social media users.

Fabian is currently pursuing his PhD at Humboldt University in the lab of Prof. Igor Sokolov where his work deals with dynamical processes on complex networks. He is especially interested in network epidemiology, random walk theory and opinion dynamics.
During his physics studies at TU Berlin he worked on mean-field methods in the context of computational neuroscience.