Opinion formation over dynamic hierarchical networks with acquaintances and strangers: A genetic variation based double-mechanism framework

Inspired by the psychological phenomenon that agents generally adopt different opinions or action-update mechanisms when faced with different types of neighbors, we propose a novel double-mechanism framework over dynamic hierarchical networks to fill this gap. First, a novel multi-attribute genetic variation-based leader-influencer-follower (LIF) dynamic hierarchical network is developed. Second, we propose a double-mechanism framework in which a synchronous asymmetric Deffuant–Weisbuch model with opinion memory effect (MSADW) and a continuous opinion and discrete action (CODA) model with action memory effect (MCODA) are built, achieving the co-evolution of the agent’s attribute and the social network with acquaintances and strangers, respectively. Finally, the experimental results show that the memory effect can effectively weaken the bounded confidence rule and behavioral preference, accelerating the consensus of group opinions or actions. In addition, all agents attempt to obtain a higher out-degree during the co-evolution process to realize the hierarchical transition. The fitted hierarchical partition functions provide a basis for adjusting social structure from the inverted T-shaped to the pyramid-shaped structure and finally to the olive-shaped structure, which has insightful social interpretations.