Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metrics
AbstractMeasuring 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...
| Main Authors: | , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
The MIT Press
2021-01-01
|
| Series: | Transactions of the Association for Computational Linguistics |
| Online Access: | https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00425/108201/Quantifying-Social-Biases-in-NLP-A-Generalization |
| _version_ | 1818441464112218112 |
|---|---|
| author | Paula Czarnowska Yogarshi Vyas Kashif Shah |
| author_facet | Paula Czarnowska Yogarshi Vyas Kashif Shah |
| author_sort | Paula Czarnowska |
| collection | DOAJ |
| description |
AbstractMeasuring 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. |
| first_indexed | 2024-12-14T18:28:40Z |
| format | Article |
| id | doaj.art-d2acc8e1928e489f8e4ff71b5ee6cc52 |
| institution | Directory Open Access Journal |
| issn | 2307-387X |
| language | English |
| last_indexed | 2024-12-14T18:28:40Z |
| publishDate | 2021-01-01 |
| publisher | The MIT Press |
| record_format | Article |
| series | Transactions of the Association for Computational Linguistics |
| spelling | doaj.art-d2acc8e1928e489f8e4ff71b5ee6cc522022-12-21T22:51:51ZengThe MIT PressTransactions of the Association for Computational Linguistics2307-387X2021-01-0191249126710.1162/tacl_a_00425Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness MetricsPaula Czarnowska0Yogarshi Vyas1Kashif Shah2University of Cambridge, UK. pjc211@cam.ac.ukAmazon AI, USA. yogarshi@amazon.comAmazon AI, USA. shahkas@amazon.com AbstractMeasuring 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.https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00425/108201/Quantifying-Social-Biases-in-NLP-A-Generalization |
| spellingShingle | Paula Czarnowska Yogarshi Vyas Kashif Shah Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metrics Transactions of the Association for Computational Linguistics |
| title | Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metrics |
| title_full | Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metrics |
| title_fullStr | Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metrics |
| title_full_unstemmed | Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metrics |
| title_short | Quantifying Social Biases in NLP: A Generalization and Empirical Comparison of Extrinsic Fairness Metrics |
| title_sort | quantifying social biases in nlp a generalization and empirical comparison of extrinsic fairness metrics |
| url | https://direct.mit.edu/tacl/article/doi/10.1162/tacl_a_00425/108201/Quantifying-Social-Biases-in-NLP-A-Generalization |
| work_keys_str_mv | AT paulaczarnowska quantifyingsocialbiasesinnlpageneralizationandempiricalcomparisonofextrinsicfairnessmetrics AT yogarshivyas quantifyingsocialbiasesinnlpageneralizationandempiricalcomparisonofextrinsicfairnessmetrics AT kashifshah quantifyingsocialbiasesinnlpageneralizationandempiricalcomparisonofextrinsicfairnessmetrics |