Rank Aggregation

Rank aggregation studies how to combine rankings or pairwise comparisons when sources disagree, are noisy, or have different levels of reliability. Classical methods include Borda-style voting, Kemeny optimization, and Markov-chain approaches, while newer methods use Bayesian models, learned rankers, and active comparison selection. Applications include search-result fusion, model and answer evaluation, recommendation, and combining human or machine judgments. A central tension is that one consensus ranking can hide genuine differences in preference, while robust, fair, and efficient aggregation becomes harder as comparisons are biased, incomplete, or computationally costly.