Research map / Learning theory and algorithms / Fairness
Breaking the Dyadic Barrier: Rethinking Fairness in Link Prediction Beyond Demographic Parity
Joao Mattos, Debolina Halder Lina, Arlei Silva
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 MORAL group-decoupled post-processing instead of other fair link prediction methods | GCN link predictor · Six fairness datasets (Facebook, Credit, German, NBA, Pokec-n, Pokec-z) | NDKL: MORAL 0.01-0.04 vs e.g. UGE 0.05-0.80 | Improved |
| use MORAL instead of DetConstSort re-ranking | GCN link predictor · Facebook, Credit prec@1000 | prec@1000: 0.00 -> 0.95 (Facebook), 0.00 -> 1.00 (Credit) | Improved |
| use MORAL instead of UGE on utility | GCN link predictor · Facebook prec@1000 | prec@1000: 0.97 -> 0.95 | No clear effect |
| enforce dyadic demographic parity instead of distribution-preserving fairness | existing fair link predictors · top-100 pair-type proportions | pair-type proportion vs target | Hurt |
| reorder ranking with the same pair proportions (worst vs best permutation) (ablation) | optimal DP ranking · controlled ranking experiment | NDKL vs delta DP | Mixed |
| add EDITS / FairAdj / FairLP / FairEGM pre- and in-processing baselines versus MORAL on large graphs | GCN · Pokec-n, Pokec-z | scalability: OOM | Hurt |
Closest papers by meaning
- Networked Inequality: Preferential Attachment Bias in Graph Neural Network Link Prediction (ICML 2024 · similarity 0.91)
- FairWire: Fair Graph Generation (NeurIPS 2024 · similarity 0.87)
- Fair Graph U-Net: A Fair Graph Learning Framework Integrating Group and Individual Awareness (AAAI 2025 · similarity 0.87)
- Arbitrariness and Social Prediction: The Confounding Role of Variance in Fair Classification (AAAI 2024 · similarity 0.86)
- Reconciling Predictive and Statistical Parity: A Causal Approach (AAAI 2024 · similarity 0.86)
- FDGen: A Fairness-Aware Graph Generation Model (ICML 2025 · similarity 0.86)
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.
