Augment opinion-dynamics models to incorporate the fine-grained impact of user-level timeline algorithms

Determine how to augment opinion-dynamics models—specifically models such as the Friedkin–Johnsen (FJ) opinion-formation model—to incorporate the fine-grained impact of user-level timeline algorithms used by online social networks, so that polarization and disagreement arising at the network level can be analyzed in a way that reflects personalized content ranking and recommendations applied at the local user level.

Background

Polarization and disagreement in online social networks are phenomena observed at the global network level, while timeline algorithms operate locally by ranking and recommending content to individual users. Traditional opinion-dynamics models, including the Friedkin–Johnsen model, typically rely on a static graph capturing social ties and do not account for recommendation-driven interactions.

Existing augmentations to the FJ model often manipulate the underlying graph directly (e.g., adding or removing a small number of edges or globally rebalancing edge weights), which does not align with how timeline algorithms actually affect content exposure and interactions. The open problem calls for a principled way to embed properties of timeline algorithms into opinion-dynamics models to bridge local personalized recommendations and global polarization metrics.