Polarisation and influence in online social networks
Recovering from Viral Narratives using Delayed Interventions and Strategic Seeding in Polarised Networks
A. Găloiu (TU Delft - Electrical Engineering, Mathematics and Computer Science)
A.L.D. Latour – Mentor (TU Delft - Electrical Engineering, Mathematics and Computer Science)
M. Khosla – Graduation committee member (TU Delft - Electrical Engineering, Mathematics and Computer Science)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Online social networks enable the rapid spread of competing narratives, including misinformation and corrective information campaigns. While previous work has studied competitive diffusion, community structure and influence maximisation separately, less attention has been given to how these factors interact when competing narratives spread according to different diffusion models.
We study the Competing Narrative Diffusion Problem, in which a viral narrative competes against a reinforcement-based intervention. We model the viral narrative using the Independent Cascade (IC) model and the corrective intervention using the Linear Threshold (LT) model. Using Stochastic Block Model and Lancichinetti-Fortunato-Radicchi networks, we investigate how community segregation, intervention timing and strategic seed placement affect diffusion outcomes.
Our results show that community structure plays a central role in determining narrative dominance. Strongly segregated networks favour LT diffusion, while increasing cross-community connectivity consistently shifts outcomes towards IC diffusion. Delayed interventions become less effective as delay increases, but remain competitive in highly segregated networks. Finally, strategic seed placement substantially improves intervention performance and allows LT diffusion to outperform IC diffusion across a much wider range of network structures.
These findings demonstrate that the effectiveness of counter-narratives depends not only on the diffusion model itself, but also on the interaction between community structure, intervention timing and seed selection. Understanding these interactions can help inform the design of more effective interventions against harmful information propagation in online social networks.