[论文] Fairness in Link Prediction Beyond Demographic Parity: A Reproducibili...

论文概要

研究领域: ML
作者: Valentijn Oldenburg, Floris de Kam, Stef de Wildt
发布时间: 2026-08-12
arXiv: 2508.05138

中文摘要

在公平的排序链接预测中，人口统计均等（\(\Delta_\mathrm{DP}\)）是常用的公平性指标。然而，Mattos等人（2025）指出它无法检测暴露偏见，因为它忽略了链接在排序中的位置。在本研究中，我们通过展示\(\Delta_\mathrm{DP}\)即使在某些子群对链接被系统性地排在更低位置时仍可能显示总体均等，来复现这一论断。然而，提出的排序感知归一化折扣KL散度（NDKL）能够检测此类差异。我们还复现了MORAL的有效性，这是一种在保持竞争性效用的同时改善基于暴露公平性的后处理方法。除了复现，我们还使用合成同质性设置、分类敏感属性以及额外的公平性和效用指标（包括子群对自适应的注意力加权排序公平性AWRF）来评估鲁棒性。总体而言，我们的结果表明，基于暴露的指标揭示了被\(\Delta_\mathrm{DP}\)隐藏的偏见，而MORAL在多样化设置和数据集上以最小效用损失减少了这些偏见。我们在https://github.com/Floris93100/reproducing-MORAL 发布了修正后的可复现实现。

原文摘要

In fair ranked link prediction, demographic parity (\(\Delta_\mathrm{DP}\)) is a common fairness metric. Yet, Mattos et al. (2025) argue that it fails to detect exposure bias because it ignores where links appear in the ranking. In this study, we reproduce this claim by showing that \(\Delta_\mathrm{DP}\) can indicate aggregate parity even when some subgroup-pair links are systematically ranked lower than others. The proposed rank-aware Normalized Discounted KL-divergence (NDKL), however, does detect such disparities. We also reproduce the effectiveness of MORAL, a post-processing method that improves exposure-based fairness while maintaining competitive utility. Beyond reproduction, we assess robustness using synthetic homophily settings, categorical sensitive attributes, and additional fairn...

自动采集于 2026-08-12

#论文 #arXiv #ML #小凯