%0 Thesis
%9 Doctoral
%A Riazi, Diana
%B Computer Science
%D 2025
%F discovery:10205241
%I UCL (University College London)
%P 125
%T Agent-based Models of Social Learning in Opinion Dynamics
%U https://discovery.ucl.ac.uk/id/eprint/10205241/
%X With phenomena such as the rise of fake news and the spread of online misinformation, the need to develop a quantitative understanding of opinion dynamics in online social networks (OSNs) has become evident. By gauging agent-based models  (ABMs), this thesis aims to better understand the relationship which exists between  mis/dis-information spread and opinion formation, in the interest of determining the  effectiveness of current counter-strategies and ultimately striving towards improved  ones.  A social learning procedure of distributed hypothesis testing (DHT) is invoked  in order to build a framework where there exists a notion of ground-truth. Agents  aim to learn this ground-truth among a set of competing hypotheses utilizing this  procedure of DHT. A sub-population of agents, referred to as conspirators which  promote a fixed non-truth, are introduced where the dynamics are observed given  a level of the network being perturbed. To understand such effects, a metric of  truthfulness is developed, which describes the network’s collective belief in the  ground-truth.  A feature of the DHT procedure invoked is the distinction between what an  agent publicly expresses and privately believes. This distinction gives rise to a  means of measuring an agent’s cognitive dissonance, evocative of the mental toll  which may arise when a decision is faced within an environment of potentially  conflicting information.  A main prong of this thesis is to understand whether the effects of misinformation may be mitigated. Given our DHT framework, it is explored whether there  are beneficial effects of the injection of noise, which has been shown to have stabilizing effects in complex system contexts. Finally, we investigate by whom or  what the moderation of misinformation should derive. We classify current counterstrategies (debunking, prebunking, deplatforming, and inoculation) as centralized  and/or decentralized and simulate such correcting interventions
%Z Copyright © The Author 2025. Original content in this thesis is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0) Licence (https://creativecommons.org/licenses/by-nc/4.0/). Any third-party copyright material present remains the property of its respective owner(s) and is licensed under its existing terms. Access may initially be restricted at the author’s request.