| We consider the problem of group opinion formation, in the presence of rapidly evolving information, in order to determine the best action to choose for influencing a human being, modeled using temporal motivational theory. In this research, we hypothesize that distributed decision making with rapidly evolving information can be done effectively if decisions are updated dynamically while incorporating the influence of different agents using opinion evolution models. To verify this hypothesis, we have developed a novel computational framework where we consider the effect of different sub-groups of people (agents) with varying degrees of influence on each agent's decision. The decisions within each sub-group are calculated at certain intervals (time-steps) using Markov decision processes. The decisions output from these sub-groups are then integrated using an opinion evolution model, which are then fed back to update the influence of each sub-group towards the decision made by this model for the next time-step. We quantitatively test the effectiveness of this model on eliciting effective actions on various state machines designed to model a potential real-world interaction. |
