Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metrics

By Paula Czarnowska, Yogarshi Vyas, Kashif Shah. 2021

140 citations recorded by the source corpus.

Abstract Measuring bias is key for better understanding and addressing unfairness in NLP/ML models. This is often done via fairness metrics, which quantify the differences in a model’s behaviour across a range of demographic groups. In this work, we shed more light on the differences and similarities between the fairness metrics used in NLP. First, we unify a broad range of existing metrics under three generalized fairness metrics, revealing the connections between them. Next, we carry out an extensive empirical comparison of existing metrics and demonstrate that the observed differences in bias measurement can be systematically explained via differences in parameter choices for our generalized metrics.

Subfields: Artificial Intelligence, Computer Science Applications, Safety Research, Natural Language Processing, Safety, Robustness & Interpretability

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Science Applications

In Safety Research

In Natural Language Processing

In Safety, Robustness & Interpretability

Citations in the corpus

“I’m sorry to hear that”: Finding New Biases in Language Models with a Holistic Descriptor Dataset (2022, background): “Adopting a general definition of bias as difference allows for NLP practitioners to make the delineation between benign and harmful for each identity term separately, based on the particular task and use case at hand (Olteanu et al., 2017; Blodgett et al., 2020; Czarnowska et al., 2021; Dev et al., ”
You reap what you sow: On the Challenges of Bias Evaluation Under Multilingual Settings (2022, background): “In response, previous work has explored ways in which such social biases can be mea-034 sured and counteracted (Nangia et al., 2020; Gehman et al., 2020; Czarnowska et al., 2021).”
Interactive Model Cards: A Human-Centered Approach to Model Documentation (2022, background): “, metrics [16, 67]), and we urge caution to ensure that these approaches are understandable and actionable for non-experts.”
Biases in Large Language Models: Origins, Inventory, and Discussion (2023, background): “Importantly, it has been found that the various sets of metrics used in hundreds of papers dealing with social bias can be uniied under three generalized fairness metrics: pairwise comparison, background comparison, and multi-group comparison metrics [41].”
The Tail Wagging the Dog: Dataset Construction Biases of Social Bias Benchmarks (2022, background): “…Nadeem et al., 2021) and expressed in specific tasks (Nangia et al., 2020; Kirk et al., 2021; Schramowski et al., 2022; Prabhumoye et al., 2021; Srinivasan and Bisk, 2021; Kirk et al., 2021; Parrish et al., 2021; Baldini et al., 2022; Czarnowska et al., 2021; Dev et al., 2021a; Zhao et al., 2021).”
Your fairness may vary: Pretrained language model fairness in toxic text classification (2021, background): “There are many deﬁnitions of group fairness and we refer to Verma and Rubin (2018) for a comprehensive overview and to Czarnowska et al. (2021) for a discussion of metrics in the context of measuring social biases in NLP. Statistical parity (SP) is one of the earlier deﬁnitions which requires the…”