Explaining Fairness-Oriented Recommendations using Transfers and Transitive Arguments - Archive ouverte HAL Accéder directement au contenu
Communication Dans Un Congrès Année : 2022

Explaining Fairness-Oriented Recommendations using Transfers and Transitive Arguments

Résumé

We explore the generation of explanations for preferences based on an ordered weighted average (OWA) model aiming to favour a balanced distribution of performances relative to the different points of view. In this paper we propose explanations, correct towards the model, based on transitivity with elementary arguments such as Pareto dominance, Pigou-Dalton transfers and the preferential information (PI) given by the decision maker. We propose several heuristic-based approaches allowing to compute such explanations, confirmed by experiments on their availability and their optimal length.
Fichier principal
Vignette du fichier
DA2PL_Explaining_Fairness_Oriented_Recommendations_using_Transfers_and_Transitive_Arguments.pdf (262.92 Ko) Télécharger le fichier
Origine : Fichiers produits par l'(les) auteur(s)

Dates et versions

hal-04310877 , version 1 (27-11-2023)

Identifiants

  • HAL Id : hal-04310877 , version 1

Citer

Hénoïk Willot, Khaled Belahcene, Sébastien Destercke. Explaining Fairness-Oriented Recommendations using Transfers and Transitive Arguments. From Multiple-Criteria Decision Aid to Preference Learning (DA2PL 2022), Khaled Belahcene; Sébastien Destercke, Nov 2022, Compiègne, France. ⟨hal-04310877⟩
12 Consultations
6 Téléchargements

Partager

Gmail Facebook X LinkedIn More