Bias Analysis and Mitigation through Protected Attribute Detection and Regard Classification
Not on the map. Why?
Abstract (arXiv)
Large language models (LLMs) acquire general linguistic knowledge from massive-scale pretraining. However, pretraining data mainly comprised of web-crawled texts contain undesirable social biases which can be perpetuated or even amplified by LLMs. In this study, we propose an efficient yet effective annotation pipeline to investigate social biases in the pretraining corpora. Our pipeline consists of protected attribute detection to identify diverse demographics, followed by regard classification to analyze the language polarity towards each attribute. Through our experiments, we demonstrate the effect of our bias analysis and mitigation measures, focusing on Common Crawl as the most representative pretraining corpus.
Key facts. Cited by 2 papers in the map (data through 14 Oct 2026). Builds on 26 of its 38 references that are in the map. First posted on arXiv 19 Apr 2025.
Builds on
26 of 38 references.
References · 26
- 2023GPT-4 Technical Reportcited by 18,396
- 2022Training language models to follow instructions with human feedbackcited by 15,971
- 2024The Llama 3 Herd of Modelscited by 14,715
- 2019RoBERTa: A Robustly Optimized BERT Pretraining Approachcited by 13,372
- 2022PaLM: Scaling Language Modeling with Pathwayscited by 5,326
- 2020RealToxicityPrompts: Evaluating Neural Toxic Degeneration in Language Modelscited by 1,255
- 2023Llama Guard: LLM-based Input-Output Safeguard for Human-AI Conversationscited by 1,207
- 2024The FineWeb Datasets: Decanting the Web for the Finest Text Data at Scalecited by 1,160
- 2023LIMA: Less Is More for Alignmentcited by 1,100
- 2019The Woman Worked as a Babysitter: On Biases in Language Generationcited by 541
- 2021Documenting Large Webtext Corpora: A Case Study on the Colossal Clean Crawled Corpuscited by 492
- 2021BOLD: Dataset and Metrics for Measuring Biases in Open-Ended Language Generationcited by 412
14 more
- 2021Self-Diagnosis and Self-Debiasing: A Proposal for Reducing Corpus-Based Bias in NLPcited by 337
- 2019Lipstick on a Pig: Debiasing Methods Cover up Systematic Gender Biases in Word Embeddings But do not Remove Themcited by 268
- 2023From Pretraining Data to Language Models to Downstream Tasks: Tracking the Trails of Political Biases Leading to Unfair NLP Modelscited by 260
- 2023The Unlocking Spell on Base LLMs: Rethinking Alignment via In-Context Learningcited by 257
- 2019Identifying and Reducing Gender Bias in Word-Level Language Modelscited by 214
- 2018Gender Bias in Neural Natural Language Processingcited by 211
- 2021Societal Biases in Language Generation: Progress and Challengescited by 202
- 2023A Pretrainer's Guide to Training Data: Measuring the Effects of Data Age, Domain Coverage, Quality, & Toxicitycited by 182
- 2021Large Pre-trained Language Models Contain Human-like Biases of What is Right and Wrong to Docited by 178
- 2021What's in the Box? A Preliminary Analysis of Undesirable Content in the Common Crawl Corpuscited by 80
- 2019GlossBERT: BERT for Word Sense Disambiguation with Gloss Knowledgecited by 79
- 2024Safeguarding Large Language Models: A Surveycited by 78
- 2023SeeGULL: A Stereotype Benchmark with Broad Geo-Cultural Coverage Leveraging Generative Modelscited by 66
- 2023ROBBIE: Robust Bias Evaluation of Large Generative Language Modelscited by 53
Built on by
Citing papers per year: 2025: 1; 2026: 1.
2 papers.