Expert Classification Aggregation

TL;DR

This paper investigates the limitations of aggregating expert classifications under certain expertise and independence conditions, revealing many impossibility results in the process.

Contribution

It introduces the concept of Expertise in classification aggregation and demonstrates the constraints and impossibility results when combined with unanimity or independence.

Findings

Expertise condition restricts aggregation outcomes.

Impossibility results under unanimity or independence.

Framework for understanding classification aggregation limitations.

Abstract

We consider the problem where a set of individuals has to classify $m$ objects into $p$ categories by aggregating the individual classifications, and no category can be left empty. An aggregator satisfies \emph{Expertise} if individuals are decisive either over the classification of a given object, or the classification into a given category. We show that requiring an aggregator to satisfy \emph{Expertise} (or variants of it) and be either unanimous or independent leads to numerous impossibility results.