Optimizing positional scoring rules for rank aggregation (Q2321258)

This paper discusses the problem of evaluating aggregating rules used to derive global rankings from incomplete rankings of a small set of alternatives by each agent. Sample sizes are assumed to be fix and constant. Real global rankings are assumed to exist, and the aggregation rules are evaluated based on the proximity to known parts of such real rankings. Positional scores are employed. Different algorithms to compute proximity are proposed and tested. Theoretic and experimental evaluations are included.