SMAA-Fair 算法伪代码
输入:决策矩阵M,群体标签G,模拟次数L
1. 标准化决策矩阵M
2. 生成L组可行权重向量{w^(l)}
3. for l = 1 to L do
4. 计算全局得分 u(w^(l)) = M w^(l)
5. 得到排名 π^(l)
6. 计算公平性指标:SP^(l), rKL^(l), nDKL^(l)
7. end for
8. 计算最差值:SP_max, rKL_max, nDKL_max
9. for l = 1 to L do
10. 对π^(l)中每个位置s的备选方案ai
11. 更新经典SMAA可接受性指数:b += 1
12. 更新公平加权的三类可接受性指数
13. end for
14. 将权重向量w^(l)存入使ai排第一的集合Wi
15. 计算公平加权中心权重向量
16. end for
17. 返回三类可接受性矩阵和三类公平中心权重向量
[1] Lahdelma, R., & Salminen, P. (2001). SMAA-2: Stochastic multicriteria acceptability analysis for group decision making. Operations Research, 49(3), 444-456.[2] Yang, K., & Stoyanovich, J. (2017). Measuring fairness in ranked outputs. In Proceedings of the 29th International Conference on Scientific and Statistical Database Management (pp. 1-6).[3] Geyik, S. C., Ambler, S., & Kenthapadi, K. (2019). Fairness-aware ranking in search & recommendation systems with application to LinkedIn Talent Search. In Proceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining (pp. 2221-2231).[4] Vetschera, R. (2017). Deriving rankings from probabilistic preferences: A survey of methods and a new approach. European Journal of Operational Research, 258(2), 629-641.[5] 原始论文: Pelegrina, G. D., & Pelissari, R. (2026). A fairness-aware extension of Stochastic Multicriteria Acceptability Analysis for ranking. arXiv:2606.17756.