Measuring Fairness with Biased Rulers: A Comparative Study on Bias Metrics for Pre-trained Language Models
Measuring Fairness with Biased Rulers: A Comparative Study on Bias Metrics for Pre-trained Language Models
复制标题
用有偏见的统治者衡量公平性：预训练语言模型的偏见度量的比较研究
DOI：
10.18653/v1/2022.naacl-main.122
复制
发表时间：
2022
期刊：
影响因子：
--
通讯作者：
Bettina Berendt
中科院分区：
文献类型：
--
作者：
Pieter Delobelle;E. Tokpo;T. Calders;Bettina Berendt
人们越来越意识到自然语言处理资源(如BERT)中的偏见模式，这促使许多指标对这些资源中的“偏见”和“公平”进行量化。然而，比较不同指标的结果和使用这些指标进行评估的工作仍然很困难，如果不是完全不可能的话。我们查阅了关于预先训练的语言模型的公平性度量的文献，并通过实验评估了兼容性，包括语言模型及其下游任务中的偏差。我们通过传统的文献调研、相关分析和实证评价相结合的方法进行研究。我们发现许多度量彼此不兼容，并且高度依赖于(I)模板、(Ii)属性和目标种子以及(Iii)嵌入的选择。我们也没有看到与外在偏见相关的内在偏见的具体证据。这些结果表明，公平或偏见评估对语境化语言模型来说仍然具有挑战性，还有其他原因，因为这些选择仍然是主观的。为了改进未来的比较和公平性评估，我们建议避免基于嵌入式的指标，并将重点放在下游任务的公平性评估上。
An increasing awareness of biased patterns in natural language processing resources such as BERT has motivated many metrics to quantify ‘bias’ and ‘fairness’ in these resources. However, comparing the results of different metrics and the works that evaluate with such metrics remains difficult, if not outright impossible. We survey the literature on fairness metrics for pre-trained language models and experimentally evaluate compatibility, including both biases in language models and in their downstream tasks. We do this by combining traditional literature survey, correlation analysis and empirical evaluations. We find that many metrics are not compatible with each other and highly depend on (i) templates, (ii) attribute and target seeds and (iii) the choice of embeddings. We also see no tangible evidence of intrinsic bias relating to extrinsic bias. These results indicate that fairness or bias evaluation remains challenging for contextualized language models, among other reasons because these choices remain subjective. To improve future comparisons and fairness evaluations, we recommend to avoid embedding-based metrics and focus on fairness evaluations in downstream tasks.
影响因子：
7.6
作者：
Greenwald, AG;McGhee, DE;Schwartz, JLK
通讯作者：
Schwartz, JLK
DOI：
10.18653/v1/n18-2003
发表时间：
2018-04
期刊：
ArXiv
影响因子：
--
作者：
Jieyu Zhao;Tianlu Wang;Mark Yatskar;Vicente Ordonez;Kai-Wei Chang
通讯作者：
Jieyu Zhao;Tianlu Wang;Mark Yatskar;Vicente Ordonez;Kai-Wei Chang