AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

Coverage, Not Averages: Semantic Stratification for Trustworthy Retrieval Evaluation

Andrew Klearman, Radu Revutchi, Rohin Garg, Rishav Chakravarti, Samuel Marc Denton, Yuan Xue

ICML 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Recommendation representation learning · 0 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use coverage-aware stratified query generation instead of original BEIR queriesNFCorpus evaluation set · NFCorpusMinimum Semantic Coverage: 50.9% -> 90.3%Improved
use coverage-aware stratified query generationNFCorpus evaluation set · NFCorpusZero-query clusters: 182 -> 36Improved
use coverage-aware stratified query generationFiQA evaluation set · FiQAZero-query clusters: 232 -> 0Improved
use coverage-aware stratified query generationSciDocs evaluation set · SciDocsMinimum Semantic Coverage: 89.2% -> 99.7%Improved
stratify queries by semantic cluster (within-stratum vs overall variance)Dense retrieval nDCG@10 · NFCorpusstandard deviation: .31 -> .29Improved
add test queries for under-covered clustersNFCorpus retrievers · NFCorpus zero-query cluster Statistical methodsnDCG@10: 0.365 / 0.336 / 0.483 (dense / BM25 / hybrid)Improved
use hybrid retrieval instead of dense or BM25 aloneStatistical methods cluster · NFCorpusnDCG@10: 0.365 -> 0.483Improved
use BM25 instead of dense retrievalCancer cell lines cluster · NFCorpusnDCG@10: 0.795 -> 0.854Improved
switch model comparison from overall mean to macro-average across clusterstext-embedding-3-small vs embed-english-v3.0 · NFCorpus, 1,000 bootstrap sampleswin rate of text-embedding-3-small: 32.2% -> 63.3%Mixed
lower relevance dispersion (Delta) at fixed alignmentQuery regime · NFCorpusmean nDCG@10: .25 -> .45 (Lo J)Improved
raise query-document semantic alignment (J) at fixed dispersionQuery regime · NFCorpusmean Recall@100: .22 -> .55 (Hi Delta)Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy