Opinion formation in multiple interconnected social systems

• Main Author: Blair, Jennifer
• Other Authors: Xiao Gaoxi
• Format: Final Year Project
• Language: English
• Published: 2019
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/78048
• Institution: Nanyang Technological University

This thesis involved designing and conducting numerical simulations on synthetic social networks to investigate the influence of various design parameters on agent opinion formation. This study is focused on community opinion formation, the largescale effect an individual both affecting and being affected by the opinions of their peers has on the opinion distribution of the whole community. Social networks consist of individuals represented as nodes and their interpersonal relationships represented as ties. As such, many social networks can be studied using techniques present in graph and network theory, however the generation and dynamics within these networks struggle to capture phenomena observed in real life. The model created throughout this project aims to combine popular opinion and network models (the Deffuant interaction model and the Barabási–Albert network model, with rewiring and mutation extensions) that generate social networks that both grow and behave like networks observed in real life. Simulations presented in several respectable papers [18,23,27] were conducted and confirmed the accuracy of the model created throughout the course of the project. With all aspects combined, the multilayer model behaviour when compared to a single layer equivalent for the same opinion threshold could produce greater or reduced opinion diversity through controlling the ratio of layer opinion thresholds. This shows that when individuals are likeminded on more than one topic, they are more likely to agree on the topic at hand, whereas distrust occurs if they have wildly different beliefs on different topics. The rewiring and mutation extensions greatly impacted the structure of the network. With high values of mutation (opinion change out with social interaction) and rewiring (ability to form or break relationships) individuals within the network could not split into isolated communities due to the frequent changes of opinion, however with low mutation probability, individuals kept true to their opinions for long periods of time and so gravitated their social connections towards likeminded groups which visibly split the network. While more work must be done to quantify the impacts of the combined model, it has been shown that the individual effects of each part of the model enhance the impact of the model as a whole, allowing the model to capture more natural social system behaviour.