Reducing COVID-19 Misinformation Spread by Introducing Information Diffusion Delay Using Agent-based Modeling

Abstract

With the explosive growth of the Coronavirus Pandemic (COVID-19), misinformation on social media has developed into a global phenomenon with widespread and detrimental societal effects. Despite recent progress and efforts in detecting COVID-19 misinformation on social media networks, this task remains challenging due to the complexity, diversity, multi-modality, and high costs of fact-checking or annotation. In this research, we introduce a systematic and multidisciplinary agent-based modeling approach to limit the spread of COVID-19 misinformation and interpret the dynamic actions of users and communities in evolutionary online (or offline) social media networks. Our model was applied to a Twitter network associated with an armed protest demonstration against the COVID-19 lockdown in Michigan state in May, 2020. We implemented a one-median problem to categorize the Twitter network into six key communities (nodes) and identified information exchange (links) within the network. We measured the response time to COVID-19 misinformation spread in the network and employed a cybernetic organizational method to monitor the Twitter network. The overall misinformation mitigation strategy was evaluated, and agents were allocated to interact with the network based on the measured response time and feedback. The proposed model prioritized the communities based on the agents response times at the operational level. It then optimized agent allocation to limit the spread of COVID19 related misinformation from different communities, improved the information diffusion delay threshold to up to 3 minutes, and ultimately enhanced the mitigation process to reduce misinformation spread across the entire network.