Lune

EMNLP Findings2022

Mind Your Bias: A Critical Review of Bias Detection Methods for Contextual Language Models

Silke Husse, Andreas Spitz

Abstract

The awareness and mitigation of biases are of fundamental importance for the fair and transparent use of contextual language models, yet they crucially depend on the accurate detection of biases as a precursor. Consequently, numerous bias detection methods have been proposed, which vary in their approach, the considered type of bias, and the data used for evaluation. However, while most detection methods are derived from the word embedding association test for static word embeddings, the reported results are heterogeneous, inconsistent, and ultimately inconclusive. To address this issue, we conduct a rigorous analysis and comparison of bias detection methods for contextual language models. Our results show that minor design and implementation decisions (or errors) have a substantial and often significant impact on the derived bias scores. Overall, we find the state of the field to be both worse than previously acknowledged due to systematic and propagated errors in implementations, yet better than anticipated since divergent results in the literature homogenize after accounting for implementation errors. Based on our findings, we conclude with a discussion of paths towards more robust and consistent bias detection methods.

Ask Lune

Know this paper in minutes, not an afternoon.

Lune can read every page of it and the top-tier papers that cite it, and gives you its claims, its evidence and its limits, each one cited to the paper it comes from. No skimming, no AI slop.

Questions to start from

Lune can draw on

the full paper · related work

Get the answer

Free to start. No credit card required.

Lineage

  • Multi-Dimensional Gender Bias Classification cites Language (Technology) is Power: A Critical Survey of "Bias" in NLP
  • Multi-Dimensional Gender Bias Classification cites CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models
  • CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models cites Multi-Dimensional Gender Bias Classification
  • Language (Technology) is Power: A Critical Survey of "Bias" in NLP cites StereoSet: Measuring stereotypical bias in pretrained language models
  • CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language Models cites StereoSet: Measuring stereotypical bias in pretrained language models
Builds onCited byLeft to right by year; the more cited, the higher.

Builds on4

  1. Language (Technology) is Power: A Critical Survey of "Bias" in NLPSu Lin Blodgett, Solon Barocas, Hal Daumé III, Hanna M. WallachACL 2020 · 68 citations
  2. CrowS-Pairs: A Challenge Dataset for Measuring Social Biases in Masked Language ModelsNikita Nangia, Clara Vania, Rasika Bhalerao, Samuel R. BowmanEMNLP 2020 · 19 citations
  3. Multi-Dimensional Gender Bias ClassificationEmily Dinan, Angela Fan, Ledell Wu, Jason Weston et al.EMNLP 2020 · 7 citations
  4. StereoSet: Measuring stereotypical bias in pretrained language modelsMoin Nadeem, Anna Bethke, Siva ReddyACL 2021

Related papers

  1. Assessing the Reliability of Word Embedding Gender Bias MeasuresYupei Du, Qixiang Fang, Dong NguyenEMNLP 2021 · 13 citations
  2. On Measuring and Mitigating Biased Inferences of Word EmbeddingsSunipa Dev, Tao Li, Jeff M. Phillips, Vivek SrikumarAAAI 2020 · 195 citations
  3. Pre-trained Speech Processing Models Contain Human-Like Biases that Propagate to Speech Emotion RecognitionIsaac Slaughter, Craig Greenberg, Reva Schwartz, Aylin CaliskanEMNLP Findings 2023 · 7 citations
  4. A Causal Inference Method for Reducing Gender Bias in Word Embedding RelationsZekun Yang, Juan FengAAAI 2020 · 40 citations
  5. Spying on Your Neighbors: Fine-grained Probing of Contextual Embeddings for Information about Surrounding WordsJosef Klafka, Allyson EttingerACL 2020 · 2 citations
Dusk over the sea between two cliffs drawn in fine vertical lines