Multi-winner Approval Voting Goes Epistemic

TL;DR

This paper explores epistemic voting for multi-winner selection, modeling votes as noisy signals about the true set of winners, and proposes rules to identify optimal winners based on prior bounds and noise models.

Contribution

It introduces a framework for multi-winner epistemic voting with bounds on the number of winners and defines optimal rules under specific noise models.

Findings

Proposed rules effectively identify true winners in noisy multi-label data.

Experimental results validate the approach on collected multi-label annotations.

Framework accommodates prior knowledge on the size of the ground truth.

Abstract

Epistemic voting interprets votes as noisy signals about a ground truth. We consider contexts where the truth consists of a set of objective winners, knowing a lower and upper bound on its cardinality. A prototypical problem for this setting is the aggre-gation of multi-label annotations with prior knowledge on the size of the ground truth. We posit noisemodels, for which we define rules that output an optimal set of winners. We report on experiments on multi-label annotations (which we collected).