Co-evolution of ideas and social networks in opinion formation

Abstract

In recent years, researches on opinion formation have attracted extensive interests and generated many pragmatic applications. Various models have been built to study the evolution of the opinion, typically based on complex social networks. The most well-known models include the Deffuant model for opinion formation with bounded confidence and a similar but not exactly the same model proposed by Hegselmann and Krause. However, these models have not thought of the truth in real life that individuals may cut off their bonds with the uncongenial neighbors holding opinions too different from theirs, and build new bonds with strangers. Based on this phenomenon, a mechanism of cutting and reconnecting is introduced into opinion formation, to make it a dynamic co-evolution process when the opinions update in the underlying network. In this dissertation, several simulations are conducted on a modified Hegselmann-Krause model in different networks, and its effect of reducing the tiny opinion clusters held by minor nodes are evaluated. Besides, a few conjectures are proposed to explain these interesting effects where the mechanism has on opinion formation and final opinion distribution.