Мультиагентна система моделювання динаміки поширення та методів нейтралізації дезінформації у соціальних мережах

Abstract

This paper presents an approach to modeling complex information processes in social media systems through developing a distributed multi-agent system. The proposed solution enables the study of the dynamics of disinformation dissemination and the effectiveness of neutralization methods under conditions of stochastic user interactions. The system implementation is based on a combination of agent-oriented modeling concepts and the analysis of cognitive parameters of network nodes. Each agent in the model is an autonomous entity with an individual level of skepticism, which determines the threshold for perceiving manipulative content. Particular attention is paid to mechanisms for actively countering disinformation through the deployment of specialized fact-checking agents capable of stopping the spread of fake news at its source and creating zones of information immunity. The introduction of controlled random variation into agent decision-making processes compensates for the uncertainty of real social communications and improves the reliability of predictive scenarios. Prototype testing confirmed the scalability of the architecture and the possibility of using it for evaluating strategies for ensuring information resilience of digital platforms. The research results provide a basis for automating the processes of detecting and mitigating disinformation attacks in real-time. The proposed approach automates the process of detecting and neutralizing manipulative content, ensuring transparency in assessing the impact of information within the network. In the long term, the developed concept can be expanded into a full-fledged adaptive system capable of self-learning based on accumulated data to counter disinformation attacks on a large scale and improve tools for monitoring users’ overall information hygiene.