Core Concepts
Study of distance functions on rankings with asymmetric treatments and distinct relevance of top and bottom positions.
Abstract
Introduction to the problem of aggregating preferences across various fields.
Importance of distance functions in evaluating cohesion and dispersion among rankings.
Shortcomings of Kendall distance in considering relative importance of positions.
Introduction of weighted top-difference distances to address these shortcomings.
Axiomatic characterization and properties of these distances for rank aggregation problems.
Application to preference aggregation, median voting rule, fairness conditions, and approximation algorithms.
Stats
The Kendall distance has two main shortcomings: not considering the relative importance of positions where swaps occur; treating all alternatives homogeneously.
Weighted top-difference distances evaluate proximity based on maximal elements, menu size, and relative importance.
Axioms A.1-A.6 provide a foundation for characterizing these distances based on betweenness axioms without neutrality requirements.
Quotes
"We introduce a class of distances that overcome shortcomings while connecting to existing metrics."
"Our distance is readily applicable to rank aggregation problems with desirable properties."