Research map / Learning theory and algorithms / Fairness
Incremental Fairness in Two-Sided Market Platforms: On Smoothly Updating Recommendations
Gourab K. Patro, Abhijnan Chakraborty, Niloy Ganguly, Krishna Gummadi
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 changed | Where | Measured | Effect |
|---|---|---|---|
| use ILP-EL rather than CanD | Incremental recommendation updates · Amazon-M, Amazon-D, GoogleLoc-F | π: 0.17 -> 0.12; 0.21 -> 0.13; 0.16 -> 0.11 | Improved |
| use ILP-EL rather than CanD | Incremental recommendation updates · Amazon-M, Amazon-D, GoogleLoc-F | Υ: 1.63 -> 1.03; 1.34 -> 1.04; 1.38 -> 1.04 | Improved |
| use ILP-EL rather than IRF | Incremental recommendation updates · Amazon-M and GoogleLoc-F | π: 0.95 -> 0.12; 0.98 -> 0.11 | Improved |
| use ILP-EL rather than IRF | Incremental recommendation updates · Amazon-M and GoogleLoc-F | Z: 0.62 -> 0.99; 0.59 -> 0.99 | Improved |
| prefilter the item space (ablation) | ILP-EL · Amazon-M, Amazon-D, GoogleLoc-F | Υ: 1.03 -> 1.16; 1.04 -> 1.45; 1.04 -> 1.33 | Hurt |
| prefilter the item space (ablation) | ILP-EL · Amazon-M, Amazon-D, GoogleLoc-F | π: 0.12 -> 0.13; 0.13 -> 0.16; 0.11 -> 0.15 | Mixed |
| use geometric rather than linear utility-threshold steps (ablation) | ILP-Estimated · Amazon-M, Amazon-D, GoogleLoc-F | π: 0.12 -> 0.28; 0.13 -> 0.22; 0.11 -> 0.19 | Hurt |
| use geometric rather than linear utility-threshold steps (ablation) | ILP-Estimated · Amazon-M, Amazon-D, GoogleLoc-F | Υ: 1.03 -> 1.18; 1.04 -> 1.12; 1.04 -> 1.16 | Hurt |
| target the previous step's observed exposure rather than estimated final exposure (ablation) | ILP with linear utility-threshold steps · Amazon-M, Amazon-D, GoogleLoc-F | π: 0.12 -> 0.46; 0.13 -> 0.74; 0.11 -> 0.29 | Hurt |
| target the previous step's observed exposure rather than estimated final exposure (ablation) | ILP with linear utility-threshold steps · Amazon-M, Amazon-D, GoogleLoc-F | Z: 0.99 -> 0.61; 0.99 -> 0.33; 0.99 -> 0.90 | Hurt |
| target the previous step's observed exposure rather than estimated final exposure (ablation) | ILP with geometric utility-threshold steps · Amazon-M, Amazon-D, GoogleLoc-F | Υ: 1.03 -> 1.10; 1.04 -> 1.08; 1.04 -> 1.24 | Mixed |
| increase the number of update steps (ablation) | ILP-EL · Amazon-M | π | Improved |
Closest papers by meaning
- Interpolating Item and User Fairness in Multi-Sided Recommendations (NeurIPS 2024 · similarity 0.86)
- Policy Design for Two-sided Platforms with Participation Dynamics (ICML 2025 · similarity 0.85)
- Fairness under Competition (NeurIPS 2025 · similarity 0.84)
- Strategic Recommendation: Revenue Optimal Matching for Online Platforms (Student Abstract) (AAAI 2024 · similarity 0.84)
- User-item fairness tradeoffs in recommendations (NeurIPS 2024 · similarity 0.84)
- FairJob: A Real-World Dataset for Fairness in Online Systems (NeurIPS 2024 · similarity 0.84)
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.
