Optimal Average Satisfaction and Extended Justified Representation in Polynomial Time
Computer Science and Game Theory
2017-04-04 v1
Abstract
In this short note, we describe an approval-based committee selection rule that admits a polynomial-time algorithm and satisfies the Extended Justified Representation (EJR) axiom. This rule is based on approximately maximizing the PAV score, by means of local search. Our proof strategy is to show that this rule provides almost optimal average satisfaction to all cohesive groups of voters, and that high average satisfaction for cohesive groups implies extended justified representation.
Keywords
Cite
@article{arxiv.1704.00293,
title = {Optimal Average Satisfaction and Extended Justified Representation in Polynomial Time},
author = {Piotr Skowron and Martin Lackner and Edith Elkind and Luis Sánchez-Fernández},
journal= {arXiv preprint arXiv:1704.00293},
year = {2017}
}