Coevolution of complex social systems and ideas : a toy model study

• Main Author: Lin, Zeyang
• Other Authors: XIAO Gaoxi
• Format: Thesis-Master by Coursework
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Science::Mathematics::Applied mathematics::Complex systems; Engineering::Electrical and electronic engineering::Computer hardware, software and systems
• Online Access: https://hdl.handle.net/10356/140981
• Institution: Nanyang Technological University

Many interaction patterns are investigated in the field of opinion dynamics and a hot direction is related to the agent’s self-properties, such as the opinion distance or character similarity. It is reasonable because people may tend to socialize with those who have close ideas with them. To discuss the phenomenon, this dissertation firstly adds a preference mechanism into the interaction agent pair selection of the original Deffuant complete mixing model and checks its consequences. Specifically, in the original Deffuant model, the interaction pair is chosen randomly at each iteration and the formation of consensus is constrained by opinion tolerance. But interactions are more likely to happen between individuals holding similar opinions. The consequent observations included: a) increased fragmentation and polarization, especially with the tolerance threshold d (0.2,0.3) , b) obvious fast convergence in lower tolerance area and c) given a large tolerance threshold, the final state is more likely to be a consensus, in this situation, the influence of preference mechanism gets weak. Besides, based on the model with the preference mechanism, this dissertation briefly discusses the rewiring operation. In general, the rewiring mechanism always shortens the neighboring link distance among the population in the early stage. Rewiring can quickly split the population in a lower opinion tolerance area and cut down the convergence time largely. Finally, some discussion on another interaction selection logic which is different from the previous cases is delivered. This logic will search and rank the connections among the population from a global view. By comparing models using different logics, the observed result is that selection from a global view seems to favor speeding up the convergence. Due to the time and hardware limitation, the newly selection logic needs to be studied more carefully with a larger population in future work.