mlsanityData to 14 Oct 2026

Flexible text generation for counterfactual fairness probing

arXiv 2206.13757 ↗First posted 28 Jun 2022cs.CL cs.CYcited by 13

Not on the map. Why?

Abstract (arXiv)

A common approach for testing fairness issues in text-based classifiers is through the use of counterfactuals: does the classifier output change if a sensitive attribute in the input is changed? Existing counterfactual generation methods typically rely on wordlists or templates, producing simple counterfactuals that don't take into account grammar, context, or subtle sensitive attribute references, and could miss issues that the wordlist creators had not considered. In this paper, we introduce a task for generating counterfactuals that overcomes these shortcomings, and demonstrate how large language models (LLMs) can be leveraged to make progress on this task. We show that this LLM-based method can produce complex counterfactuals that existing methods cannot, comparing the performance of various counterfactual generation methods on the Civil Comments dataset and showing their value in evaluating a toxicity classifier.

Key facts. Cited by 13 papers in the map (data through 14 Oct 2026). Builds on 33 of its 53 references that are in the map. First posted on arXiv 28 Jun 2022.

Builds on

33 of 53 references.

References · 33

  1. 2018BERT: Pre-training of Deep Bidirectional Transformers for Language Understandingcited by 40,399
  2. 2020Language Models are Few-Shot Learnerscited by 32,419
  3. 2019BERTScore: Evaluating Text Generation with BERTcited by 5,512
  4. 2021On the Opportunities and Risks of Foundation Modelscited by 4,033
  5. 2021The Power of Scale for Parameter-Efficient Prompt Tuningcited by 3,795
  6. 2021Finetuned Language Models Are Zero-Shot Learnerscited by 3,780
  7. 2022LaMDA: Language Models for Dialog Applicationscited by 1,220
  8. 2019Plug and Play Language Models: A Simple Approach to Controlled Text Generationcited by 799
  9. 2019Universal Adversarial Triggers for Attacking and Analyzing NLPcited by 722
  10. 2019The Woman Worked as a Babysitter: On Biases in Language Generationcited by 541
  11. 2018Gender Bias in Coreference Resolutioncited by 442
  12. 2019Learning the Difference that Makes a Difference with Counterfactually-Augmented Datacited by 439
21 more
  1. 2019Nuanced Metrics for Measuring Unintended Bias with Real Data for Text Classificationcited by 404
  2. 2019Social Bias Frames: Reasoning about Social and Power Implications of Languagecited by 380
  3. 2021Persistent Anti-Muslim Bias in Large Language Modelscited by 377
  4. 2021Polyjuice: Generating Counterfactuals for Explaining, Evaluating, and Improving Modelscited by 222
  5. 2020HateCheck: Functional Tests for Hate Speech Detection Modelscited by 216
  6. 2019Racial Bias in Hate Speech and Abusive Language Detection Datasetscited by 209
  7. 2020Social Biases in NLP Models as Barriers for Persons with Disabilitiescited by 182
  8. 2019Queens are Powerful too: Mitigating Gender Bias in Dialogue Generationcited by 152
  9. 2018Counterfactual Fairness in Text Classification through Robustnesscited by 151
  10. 2020Towards Controllable Biases in Language Generationcited by 112
  11. 2021A Recipe For Arbitrary Text Style Transfer with Large Language Modelscited by 107
  12. 2020Explaining NLP Models via Minimal Contrastive Editing (MiCE)cited by 106
  13. 2019It's All in the Name: Mitigating Gender Bias with Name-Based Counterfactual Data Substitutioncited by 103
  14. 2021Challenges in Automated Debiasing for Toxic Language Detectioncited by 96
  15. 2020Generate Your Counterfactuals: Towards Controlled Counterfactual Generation for Textcited by 75
  16. 2019Perturbation Sensitivity Analysis to Detect Unintended Model Biasescited by 74
  17. 2021Tailor: Generating and Perturbing Text with Semantic Controlscited by 68
  18. 2021Hatemoji: A Test Suite and Adversarially-Generated Dataset for Benchmarking and Detecting Emoji-based Hatecited by 50
  19. 2020CAT-Gen: Improving Robustness in NLP Models via Controlled Adversarial Text Generationcited by 44
  20. 2021Improving Counterfactual Generation for Fair Hate Speech Detectioncited by 10
  21. 2022Counterfactual Multi-Token Fairness in Text Classificationcited by 3

Built on by

Cited by year

Citing papers per year: 2023: 8; 2024: 5.

13 papers.

Sort:

By area:

  1. 2023AART: AI-Assisted Red-Teaming with Diverse Data Generation for New LLM-powered Applicationscited by 54
  2. 2023SocialCounterfactuals: Probing and Mitigating Intersectional Social Biases in Vision-Language Models with Counterfactual Examplescited by 53
  3. 2024Navigating LLM Ethics: Advancements, Challenges, and Future Directionscited by 28
  4. 2023CATfOOD: Counterfactual Augmented Training for Improving Out-of-Domain Performance and Calibrationcited by 17
  5. 2024Is Your Large Language Model Knowledgeable or a Choices-Only Cheater?cited by 12
  6. 2024RATE: Causal Explainability of Reward Models with Imperfect Counterfactualscited by 8
  7. 2023Visualizing Linguistic Diversity of Text Datasets Synthesized by Large Language Modelscited by 4
  8. 2024Language Models as Causal Effect Generatorscited by 4
  9. 2023MLHOps: Machine Learning for Healthcare Operationscited by 2
  10. 2023On Bias and Fairness in NLP: Investigating the Impact of Bias and Debiasing in Language Models on the Fairness of Toxicity Detectioncited by 1
  11. 2023Synthetic Text Generation using Hypergraph Representationscited by 1
  12. 2024Causal Inference for Human-Language Model Collaborationcited by 1
1 more
  1. 2023Measuring and Improving Attentiveness to Partial Inputs with Counterfactualscited by 0