Bayesian inference and ant colony optimization for multi-rumor mitigation in online social networks

With the increasing popularity of social media, Online Social Networks (OSNs) are being used for promoting or discrediting various products or persons. As such, rumors are spread in the networks to increase or decrease the popularity of the target. Limiting the spread of rumors is an important research problem. In a promotion or smear campaign, we see multiple rumors about the target. Many existing works have explored rumor propagation and mitigation in social networks for a single rumor. However, users become biased towards the topic due to multiple rumors about it. A user influenced by the previous rumors about a topic is more likely to believe a rumor with similar content. Therefore, in this work, we analyze the spread of multiple rumors about a topic and formulate an optimization problem to identify the top k rumor spreaders. A Bayesian Inference has been applied to model the user bias caused by multiple rumors based on rumor content and user opinion about the topic. An Adaptive Ant Colony Optimization algorithm has been proposed to determine the top k rumor spreaders, who may be isolated from the network to reduce the impact of the rumors in the OSN. The efficacy of the proposed approaches is shown through experimentation on two datasets by considering the budget k.