- FR
- EN
Abstract
Traditionally in social choice it is supposed that the different components (candidates, preferences, and the voting rule) of the decision problem are precisely known and fixed beforehand. However, in modern applications of social choice, where choices are often made with the aid of computer tools, this assumption is not necessarily reasonable.
One source of uncertainty concerns the candidates: it is possible that some candidates may become unavailable after the voters has expressed their preferences. The second source of uncertainty concerns the preferences of the voters:requiring agents to express full preference orderings can be prohibitively costly (in terms of cognitive and communication cost) and therefore we may need to deal with situations where only partial preference information is available.
Finally, the chair (the person or organization supervising the voting process) may not be able to provide a precise definition of the voting rule. While non experts may not be able to formalize a voting rule, they might provide information about the voting rule by answering a few questions requiring to select who should be the winner out of specific synthetic profiles.
In this talk I will present some contributions concerning these three types of uncertainty in social choice. In particular, I will present an extension of Lu and Boutilier’s «unvailable candidate model» for dealing with candidates whose availability is uncertain, aggregation methods considering several plausible social choice rules given a set of desirable properties, and approaches for dealing with partially specified preferences and as well elicit voters preference, mentioning as well the possibility of integrating different types of uncertainty.
*References*:
Paolo Viappiani. Robust winner determination in positional scoring rules with uncertain weights. Theory & Decision, 88, pages 323–367, 2020
Arnaud Grivet Sébert, Nicolas Maudet, Patrice Perny, Paolo Viappiani: Rank Aggregation by Dissatisfaction Minimisation in the Unavailable Candidate Model. (To appear in the proceedings of ADT 2021)
Beatrice Napolitano, Olivier Cailloux and Paolo Viappiani. Simultaneous Elicitation of Scoring Rule and Agent Preferences for Robust Winner Determination. (To appear in the proceedings of ADT 2021)
