Dynamics of opinion formation on complex social networks

• Main Author: Nguyen, Xuan Vu
• Other Authors: XIAO Gaoxi
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Electrical and electronic engineering; Engineering::Computer science and engineering
• Online Access: https://hdl.handle.net/10356/143961
• Institution: Nanyang Technological University

Complexity science is an emerging interdisciplinary research field that has been grabbing a great deal of attention over the past few decades. The field mainly studies collective dynamics of complex systems and networks emerging from interactions among their interconnected components. Such complex systems and networks are known to be ubiquitous in a huge variety of areas ranging from biology, neurology, health, and medicine to sociology, economics, and communication networks, etc. The rise of modern network sciences and the advances in studies on dynamical processes on complex networks and systems have provided the insightful understanding and novel approaches to computational sociology. Such studies pose interesting questions whose solutions may make impacts on economics, sociology, and politics, etc. By basing opinion dynamics on mathematical models, the existing studies, to a significant extent, have exhibited observations resembling real-life phenomena and revealed key factors playing important roles in the construction of social structures and the formation of public opinions. In this thesis, four problems on dynamics of opinion formation have been investigated which, to the best of our knowledge, have been largely missed in existing studies. The four problems include: 1. In the modern world, people may be active on multiple social networks such as Reddit, Twitter or Facebook. As a result, single social networks have evolved to multiplex networks. Understanding the dynamics of opinion formation on multiplex networks is of both research interest and application values. We considered the rules proposed in the bounded confidence opinion dynamics models and examine the effects of the interplay between layers of a multiplex network on the evolution of public opinions governed by such rules. The results show that the interactions of individuals on multiple different layers of a multiplex network may diminish or enhance the opinion diversity depending on the tolerance threshold of each layer. 2. Majority rule is one of the most popular tendencies in human behaviors in choosing either of two alternatives. Following the classical majority rule, one tends to adopt the opinion shared by the majority of the connections s/he has. Under such a regime, it is shown in most of the existing studies that a complete consensus is achieved across the population in relatively dense networks. However, we figure out that in sparse networks, the dynamics may be very different where multiple steady states of co-existence could emerge. Moreover, we examine a modified majority rule where different social influences of different individuals are taken into account. It is revealed that under such a rule, once again multi-steady state of coexistence could emerge in sparse networks, yet due to very different reasons from those for the case under the classic majority rule. 3. Classical bounded-confidence models mostly deal with pairwise interactions where the two involving agents have equal influences on each other. In real life, human contacts may be more complex and sophisticated. We examine opinion dynamics under the effects of bias in social interactions. Theoretical and simulation results show that the unbalance in interpersonal contacts may lead to macroscopic polarization and/or the emergence of extremism in the entire opinion system. Influences of a few other factors on the emergence and prevalence of extremism are also discussed. 4. Most of existing opinion dynamics models aim to reveal influences of a certain key factor in opinion formation. Dynamics of opinion formation under the interplay of multiple social rules and principles are largely unknown. We examine a simple model where individuals interact with each other under the influences of the two rules of interpersonal consensus making and majority orientation. We show that some interesting and complex system dynamics shall then emerge. Our contributions as listed above shall help provide deeper insights into opinion formation and evolution in human societies. A few directions for future research are also discussed.