Pith. sign in

REVIEW 2 cited by

Efficient computation of rankings from pairwise comparisons

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2207.00076 v2 pith:RSRKHVOX submitted 2022-06-30 stat.ML cs.LG

Efficient computation of rankings from pairwise comparisons

classification stat.ML cs.LG
keywords algorithmcomparisonsfastermodelpairwiserankingsresultssimple
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
read the original abstract

We study the ranking of individuals, teams, or objects, based on pairwise comparisons between them, using the Bradley-Terry model. Estimates of rankings within this model are commonly made using a simple iterative algorithm first introduced by Zermelo almost a century ago. Here we describe an alternative and similarly simple iteration that provably returns identical results but does so much faster -- over a hundred times faster in some cases. We demonstrate this algorithm with applications to a range of example data sets and derive a number of results regarding its convergence.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Ordering Stochastic Block Models via prior transitivity

    stat.ME 2026-06 unverdicted novelty 7.0

    Introduces TSBM, a new Bayesian model for directed networks that enforces ordered blocks via transitivity-inducing priors on directional imbalance and jointly infers block count with an age-ordered partition prior.

  2. Pairwise Reference Alignment as a Model-Level Ordinal Observable

    cs.CL 2026-05 unverdicted novelty 5.0

    Pairwise reference alignment is formulated as an ordinal observable equal to the probability that a model score agrees with reference preferences on triples (x, y+, y-), with centered statistics, margin extensions, es...