Challenges in Automated Debiasing for Toxic Language Detection

Abstract

Biased associations have been a challenge in the development of classifiers for detecting toxic language, hindering both fairness and accuracy. As potential solutions, we investigate recently introduced debiasing methods for text classification datasets and models, as applied to toxic language detection. Our focus is on lexical (e.g., swear words, slurs, identity mentions) and dialectal markers (specifically African American English). Our comprehensive experiments establish that existing methods are limited in their ability to prevent biased behavior in current toxicity detectors. We then propose an automatic, dialect-aware data correction method, as a proof-of-concept. Despite the use of synthetic labels, this method reduces dialectal associations with toxicity. Overall, our findings show that debiasing a model trained on biased toxic language data is not as effective as simply relabeling the data to remove existing biases.

EACL 2021

References in corpus (5)

• Equality of Opportunity in Supervised Learning
• Hate Speech Detection and Racial Bias Mitigation in Social Media based on BERT model
• Measuring the Reliability of Hate Speech Annotations: The Case of the European Refugee Crisis
• Adversarial Filters of Dataset Biases
• Intersectional Bias in Hate Speech and Abusive Language Datasets

Cited by in corpus (6)

• Confronting Abusive Language Online: A Survey from the Ethical and Human Rights Perspective
• General-Purpose Question-Answering with Macaw
• LMMS Reloaded: Transformer-based Sense Embeddings for Disambiguation and Beyond
• Anticipating Safety Issues in E2E Conversational AI: Framework and Tooling
• On a Benefit of Mask Language Modeling: Robustness to Simplicity Bias
• #ContextMatters: Advantages and Limitations of Using Machine Learning to Support Women in Politics