A Behavioral Assessment Model for Emotional Persuasion Driven by Agent-Based Decision-Making

Abstract

Emotional persuasion driven by agent-based decision-making has shown greater prospects for negotiation between business partners. However, the interactive assessment of emotional persuasion, one of the essential preconditions for formulating an appropriate proposal, is still in its infancy. To that end, this research proposes a modeling framework that integrates three types of assessment behavior: emotion, time belief, and the evaluation of the opponent’s concession behavior. Based on the Weber-Fechner’s law, we designed a method to calculate an agent’s emotional assessment of the opponent’s proposal. Next, we construct a time belief function which can dynamically assess the pace of negotiation. Then, a Q-Learning algorithm is used to integrate those two assessments with evaluating another agent’s concession behavior to update the focal agent’s proposal. Moreover, a series of numerical experiments are conducted. The results combined with sensitivity and comparative analyses show that the proposed model framework is superior to the existing agent-based persuasion methods in terms of negotiation efficiency and fairness.