Modelling Irrational Agent Beliefs In Online Social Networks

The spread of misinformation through online social network platforms have become a major concern in society.

Understanding human behaviour and decision-making in complex systems requires modelling irrational beliefs of actors in social networks. Irrational beliefs can drive people to make decisions that are counter to their own interests or the greater good, producing outcomes that are less than ideal for both the individual and society. This thesis addresses the problem of modelling irrational beliefs in social networks by creating a framework that reflects the impact of such beliefs on agent behaviour. Graph neural networks are increasingly employed to model how beliefs propagate across a network of interconnected agents and to explore how they affect outcomes in a social system.

This research presents a comprehensive review of the latest advancements in the use of graph neural networks for the purpose of modelling irrational agent beliefs in social networks. The approach represents agents and their interactions as nodes and edges in a graph. GNNs’ are then used to learn the underlying structure and dynamics of the network, with a focus on understanding how irrational beliefs propagate through the network. The proposed framework incorporates the effects of social influence and biases into a GNN model of agent behaviour and is intended to provide insights into how misinformation and other forms of irrationality can spread within social networks and may have implications for understanding and mitigating the effects of disinformation and other forms of misinformation.