A Scalable Opinion Dynamics Model Based on the Markovian Agent Paradigm

Agent-based approaches are frequently used in the literature for the analytical modeling of opinion dynamics. In the cases in which each agent is modeled through a Markov chain, state-space explosion arises as main limit to the scalability of the model. Lumping or simulation are often the only alternatives. In this paper, we propose the use of Markovian Agents to model Peer Assembly scenarios in which agents influence each other in an atomic and linear way. Results show how this modeling paradigm is effective and flexible in providing transient and steady-state solutions for several conditions of parameter settings. This opens the way for future work in modeling more complex scenarios such as gossip-based interactions or the presence of stubborn agents.