Developing Epidemiological Models with Differentiated Infected Intensity

This study investigates the spread of toxic content on social media, a growing concern as online platforms serve as primary information sources. This research aims to enhance the accuracy of models capturing toxicity spread by differentiating between varying levels of toxicity intensity. Two epidemiological models are developed and assessed: the SEIRS model and a novel \(SEI_{m}I_{h}RS\) model. The latter divides infected users into moderate and highly infected groups to reflect the varying severity of toxic behavior. Both models are tested on six datasets to evaluate their performance. The \(SEI_{m}I_{h}RS\) model achieves even lower error rates, indicating a more precise representation of toxicity propagation. This research contributes a sophisticated tool for analyzing online toxicity, aiding policymakers and online platforms in developing targeted interventions and enhancing content moderation systems.