Minimizing the misinformation concern over social networks

Abstract

People’s opinions are often affected by their social network, and the associated misinformation on the online social networks can easily mislead people’s judgment and decision-making process, leading people to take unconventional or even radical behaviors. People’s decision-making behavior is influenced by their concern to the misinformation they receive. Building on this, we explore the competitive concern minimization problem of leveraging agents who post correct information to minimize users’ concern to misinformation. First, considering users’ concern to misinformation, this paper constructs a concern-critical competitive model and introduces the Coulomb’s law to quantify the dynamic evolution of users’ concern in information diffusion. Second, we prove hardness results for the competitive concern minimization problem and discuss the modularity of the objective function. Then, to optimize the non-submodular objective function, a two-stage approximate projected subgradient algorithm with data-dependent approximation ratio is developed using Lovász extension and convex envelope. Finally, the experimental simulations on three real networks highlight the efficiency of the approaches proposed in this paper, which is at least 9.71% better than other baselines in reducing misinformation concern.