Minimizing Polarization and Disagreement in Social Networks
This addresses the challenge of mitigating societal polarization and disagreement in online social networks, which is crucial for recommender systems and social media platforms, though it is incremental in applying optimization techniques to opinion dynamics.
The paper tackles the problem of designing social network structures to minimize both polarization and disagreement among agents with initial opinions, formalizing it as an optimization problem and providing an exact algorithm with approximation guarantees. It achieves a significant reduction, such as a ~60,000-fold decrease in polarization and disagreement on a Reddit politics network.
The rise of social media and online social networks has been a disruptive force in society. Opinions are increasingly shaped by interactions on online social media, and social phenomena including disagreement and polarization are now tightly woven into everyday life. In this work we initiate the study of the following question: given $n$ agents, each with its own initial opinion that reflects its core value on a topic, and an opinion dynamics model, what is the structure of a social network that minimizes {\em polarization} and {\em disagreement} simultaneously? This question is central to recommender systems: should a recommender system prefer a link suggestion between two online users with similar mindsets in order to keep disagreement low, or between two users with different opinions in order to expose each to the other's viewpoint of the world, and decrease overall levels of polarization? Our contributions include a mathematical formalization of this question as an optimization problem and an exact, time-efficient algorithm. We also prove that there always exists a network with $O(n/ε^2)$ edges that is a $(1+ε)$ approximation to the optimum. For a fixed graph, we additionally show how to optimize our objective function over the agents' innate opinions in polynomial time. We perform an empirical study of our proposed methods on synthetic and real-world data that verify their value as mining tools to better understand the trade-off between of disagreement and polarization. We find that there is a lot of space to reduce both polarization and disagreement in real-world networks; for instance, on a Reddit network where users exchange comments on politics, our methods achieve a $\sim 60\,000$-fold reduction in polarization and disagreement.