Fair and Explainable Collective Decision

Many real-life problems involve making a collective decision. One can cite, e.g., the choice of projects to fund in cities, or the assignment of students to courses or universities (see, e.g., in France the Parcoursup application for graduate studies). The former problem is known as participatory budgeting, a participatory democratic approach which is increasingly adopted in many countries around the world. The latter falls into the class of matching problems under preferences, where agents need to be matched to elements (tasks, roommates, jobs, etc.) based on their preferences.

In such concrete and societal problems, it is crucial to guarantee that the algorithms used to compute the collective decision are fair to the agents, in order to ensure confidence and participation in the system. The guarantee of fairness in collective decision making can be achieved via the justification of the final decision to satisfy a given appropriate fairness concept. This involves designing realistic and achievable fairness concepts, but also to explain that the final outcome is actually fair.

Important research questions remain on the design of adequate notions of fairness and their explanation for collective decision problems. The goal of this workshop is to bring together researchers from different communities in order to share insights on these topics and to progress towards a better understanding of fairness and its explanation in collective decision making. The focus will be on two specific topics: participatory budgeting and matchings under preferences. The main research questions we can investigate in these two settings are the following:

• What are adequate definitions of fairness, and how fair solutions can be efficiently computed?
• What are the characteristics of a relevant explanation for explaining fairness?
• How to design algorithmically efficient methods/approaches to compute explanations for fair decisions?
• How to evaluate the relevance of explanations that would make these explanations acceptable/understandable/relevant for end-users (metrics, methodologies, etc.)?
• How to integrate the possibility for end-users to challenge/question the result of a fair collective decision and obtain a reasoning to justify/support such a decision?

Local organizing committee

• Chloé Le Bail (LISN, Université Paris-Saclay)
• Vincent Mousseau (MICS, CentraleSupélec, Université Paris-Saclay)
• Wassila Ouerdane (MICS, CentraleSupélec, Université Paris-Saclay)
• Fabien Tarissan (CNRS, ENS Paris-Saclay)
• Anaëlle Wilczynski (MICS, CentraleSupélec, Université Paris-Saclay)