The Inverse GMCR (Graph Model for Conflict Resolution) produces rankings of possible states (preference relation profiles) that will make the desired resolution of a conflict stable. However, there are usually numerous preference relation profiles making it difficult for a third party to choose an appropriate preference relation to design its mediation strategy. Moreover, the cost or effort of changing preference relations over states has rarely been studied in Inverse GMCR. The current study presents two inverse preference optimization models considering the cost and effort in changing preferences to address these issues. The first model aims to ascertain an optimal preference at minimum adjustment cost such that the desired equilibrium is reached. The other model is to find an optimal required preference under minimum adjustment amount, which is defined as the difference between the required preference matrix and the original preference matrix. Then, a Genetic Algorithm (GA)-based algorithm is proposed. Finally, the two proposed preference optimization methods are applied to two cases, demonstrating the effectiveness of the proposed methodology.

逆 GMCR（冲突解决图形模型）生成可能状态（偏好关系配置文件）的排名，这将使所需的冲突解决方案稳定。然而，通常有许多偏好关系配置文件，这使得第三方难以选择合适的偏好关系来设计其调解策略。此外，在逆 GMCR 中很少研究改变对状态的偏好关系的成本或努力。当前的研究提出了两种逆偏好优化模型，考虑到改变偏好以解决这些问题的成本和努力。第一个模型旨在以最小的调整成本确定最佳偏好，从而达到所需的平衡。另一种模型是在最小调整量下找到最优的所需偏好，它被定义为所需偏好矩阵与原始偏好矩阵之间的差异。然后，提出了一种基于遗传算法（GA）的算法。最后，将所提出的两种偏好优化方法应用于两种情况，证明了所提出方法的有效性。