Assessing strategies for controlling viral rumor propagation on social media - a simulation approach

Like most cyber-networks, the increasing adoption of social media for a multitude of communication purposes makes them an attractive target for hacking and rumormongering. In this work, two specific strategies to control harmful viral rumor activity are explored via extensive simulations based on probabilistic models. A micro-scale generative model for context-driven viral activity is first presented and analyzed. The first strategy is formulated as the inverse of seeding influential users to achieve maximal content spread. Activity-based centrality computation is explored as the strategy to target context-dependent influential nodes whose messages are probabilistically dropped. The second strategy makes use of a softer approach in which specific agents designated to counter the rumors do so by spreading anti-rumor messages. We also show via simulations that in the event of hacking of social media accounts of highly influential nodes, the first strategy presents a better alternative in controlling the viral rumors.