A Nearly-Linear Time Algorithm for Minimizing Risk of Conflict in Social Networks

TL;DR

This paper introduces a nearly-linear time algorithm for minimizing conflict risk in social networks by modifying opinions, achieving high efficiency and effectiveness on large-scale datasets with theoretical guarantees.

Contribution

It presents a nearly-linear time algorithm with approximation guarantees for conflict minimization, improving scalability over existing methods.

Findings

Algorithm achieves up to 20x speed-up on large networks.

Effective in reducing conflict risk with high approximation accuracy.

Scales to networks with over two million nodes.

Abstract

Concomitant with the tremendous prevalence of online social media platforms, the interactions among individuals are unprecedentedly enhanced. People are free to interact with acquaintances, express and exchange their own opinions through commenting, liking, retweeting on online social media, leading to resistance, controversy and other important phenomena over controversial social issues, which have been the subject of many recent works. In this paper, we study the problem of minimizing risk of conflict in social networks by modifying the initial opinions of a small number of nodes. We show that the objective function of the combinatorial optimization problem is monotone and supermodular. We then propose a na\"{\i}ve greedy algorithm with a $(1-1/e)$ approximation ratio that solves the problem in cubic time. To overcome the computation challenge for large networks, we further integrate several effective approximation strategies to provide a nearly linear time algorithm with a $(1-1/e-\epsilon)$ approximation ratio for any error parameter $\epsilon>0$. Extensive experiments on various real-world datasets demonstrate both the efficiency and effectiveness of our algorithms. In particular, the fast one scales to large networks with more than two million nodes, and achieves up to $20\times$ speed-up over the state-of-the-art algorithm.