Modeling and simulation of interventions’ effect on the spread of toxicity in social media

The prevalence of toxicity on social media platforms constitutes a significant issue. Gaining insights into the factors that contribute to toxicity is essential for devising effective strategies to mitigate it. In this work, we extend and evaluate fractional control SEIQR (Susceptible, Exposed, Infected, Quarantined, Recovered) epidemiological modeling incorporating government interventions and awareness programs. The model incorporates different infected groups, moderate and high infected users, and is used to investigate the influence by each user on the overall spread of toxicity. We have evaluated the toxic post-free equilibrium point, the reproduction number $\left(R_0\right)$, the existence-uniqueness, and the stability point. We performed the model sensitivity analysis using the Latin Hypercube Sampling-Partial Rank Correlation Coefficient (LHS-PRCC) method. For data fitting analysis, we examined data from COVID-19-related tweets. We examine the intricacies of the proposed numerical scheme, providing a robust framework for analyzing and comprehending online toxicity spread. Simulations were conducted to elucidate the effects of government interventions and public awareness programs on the prevalence and dynamics of online toxicity spread. The study’s primary accomplishment is the model’s reduction of the error rate to 0.0011. This is distinguished by the reduced need to remove users from the network. The model not only improves accuracy but also maintains a larger user base, indicating an efficient, user-centric strategy. The results suggest that both awareness programs and government interventions are crucial for managing and mitigating online toxicity spread. This study will significantly assist network providers and policymakers to identify the infected users, thereby reducing toxic conversations.