Research map / Learning theory and algorithms / Fairness
On the Problem of Underranking in Group-Fair Ranking
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 changed | Where | Measured | Effect |
|---|---|---|---|
| Increase the minimum protected-group representation requirement | Group-fair re-ranking on COMPAS, German Credit Risk, and three-group age settings · Top 20, 40, and 100 ranks | underranking: -0.15 to 0.2 | Hurt |
| Use ALG instead of Celis et al.'s DP algorithm | Group-fair re-ranking on the real-world datasets · Top-k representation and underranking | underranking | Helped |
| Use ALG instead of FA*IR | Group-fair re-ranking on the real-world datasets · Top-k representation and underranking | representation | Helped |
| Use ALG instead of Celis et al.'s DP algorithm | German Credit Risk, age <25 · k=100 | average wall-clock running time (seconds): 11.0 vs 1.1 seconds | Helped |
| Use ALG instead of FA*IR | German Credit Risk, age <25 · k=100 | average wall-clock running time (seconds): 3.0 vs 1.1 seconds | Helped |
| Use group-fair re-ranking instead of the true ranking when the protected group is underrepresented near the top | Protected groups in the COMPAS and German Credit Risk datasets · Top-k ranks | underranking | Hurt |
| Evaluate underranking in addition to nDCG and group fairness | Comparing group-fair rankings · Top-k rankings | underranking | Helped |
Closest papers by meaning
- Fair Rank Aggregation (NeurIPS 2022 · similarity 0.91)
- Fairness in Aggregation: Optimal Top-$k$ and Improved Full Ranking (ICML 2026 · similarity 0.90)
- Fair Ranking with Noisy Protected Attributes (NeurIPS 2022 · similarity 0.89)
- Stability and Multigroup Fairness in Ranking with Uncertain Predictions (ICML 2024 · similarity 0.87)
- Understanding challenges to the interpretation of disaggregated evaluations of algorithmic fairness (NeurIPS 2025 · similarity 0.84)
- Centralized Selection with Preferences in the Presence of Biases (ICML 2024 · similarity 0.83)
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.