We consider the discrete assignment problem in which agents express ordinal preferences over objects and these objects are allocated to the agents in a fair manner. We use the stochastic dominance relation between fractional or randomized allocations to systematically define varying notions of proportionality and envy-freeness for discrete assignments. The computational complexity of checking whether a fair assignment exists is studied for these fairness notions. We also characterize the conditions under which a fair assignment is guaranteed to exist. For a number of fairness concepts, polynomial-time algorithms are presented to check whether a fair assignment exists. Our algorithmic results also extend to the case of unequal entitlements of agents. Our NP-hardness result, which holds for several variants of envy-freeness, answers an open question posed by Bouveret, Endriss, and Lang (ECAI 2010). We also propose fairness concepts that always suggest a non-empty set of assignments with meaningful fairness properties. Among these concepts, optimal proportionality and optimal weak proportionality appear to be desirable fairness concepts.
翻译:我们考虑不同的分配问题,即代理人表示对对象的偏好,这些对象被公平地分配给代理人。我们使用分数分配或随机分配之间的随机支配关系,系统地界定不同相称性和不嫉妒性的概念。为了这些公平概念,我们研究了检查是否存在公平转让的计算复杂性。我们还界定了保证公平转让存在的条件。为了一些公平概念,我们提出多时算法,以检查是否存在公平转让。我们的算法结果还涉及代理人权利不平等的情况。我们的NP-硬性结果,它包含多种无嫉妒性,回答了布韦雷特、恩德里斯和兰格(ECAI,2010年)提出的一个开放问题。我们还提出了公平概念,这些概念总是提出一套具有有意义的公平属性的、非空白性的分配。这些概念中,最佳的相称性和最弱的相称性似乎是可取的公平概念。