Active strategy learning-based consensus mechanism for group decision making with efficiency and fairness

Efficiency and fairness constitute pivotal criteria for evaluating consensus mechanisms in group decision-making (GDM) systems, where both active strategy learning and dynamic weight allocation critically impact these metrics. Active strategy learning empowers decision makers (DMs/agents) to iteratively optimize consensus strategies through Q-learning-based environmental interactions, effectively balancing individual preferences with collective objectives. Concurrently, a contribution-weighted allocation mechanism ensures fairness while maintaining original opinion authenticity. To address existing research gaps in integrated learning frameworks, fairness quantification, and decision rationality verification, this study proposes an adaptive Q-learning consensus model enhancing efficiency through reward feedback mechanisms. A dual-factor reward function integrating contribution weights and opinion retention to improve fairness metrics. An algorithmic framework with complexity analysis is validated through multi-scenario simulations. Empirical results from case study and simulation analysis demonstrate the method’s superior performance in achieving efficient convergence, equitable resource distribution, and rational opinion evolution compared to conventional approaches.