Elamur Research⌕
☰
home / papers / 2021 / Collecting a Large-Scale Gender Bias Dataset for Coreference Resolution and Machine Translation

Collecting a Large-Scale Gender Bias Dataset for Coreference Resolution and Machine Translation

Summary

Earlier studies of gender bias in coreference resolution and machine translation relied on small, template-built test sets such as Winogender and WinoBias. These sentences are artificial and unlike typical training text, and the sets are too small to train on. The authors built BUG, a corpus of about 108K naturally occurring English sentences. Each sentence links a profession noun to a gendered pronoun. They wrote 14 lexical-syntactic patterns based on the synthetic benchmarks and ran them with the SPIKE search engine over Wikipedia, PubMed abstracts and Covid19 research papers. They then labelled each match as stereotypical, anti-stereotypical or neutral using U.S. census statistics on occupational gender.

All tested translation and coreference systems handled feminine and anti-stereotypical cases worse. This shows that stereotype reliance persists on realistic input, not only on contrived examples. Fine-tuning a coreference model on the anti-stereotypical part of BUG roughly halved its bias-related errors on held-out data, with only a small loss in overall accuracy. BUG can therefore serve both to measure bias and to reduce it.

Key points

  • Size and sources: BUG has 108,419 sentences: about 80K from Wikipedia, 24K from PubMed abstracts and 4K from Covid19 research.
    • By gender-role label: 54,009 stereotypical, 29,877 anti-stereotypical and 24,533 neutral.
    • Masculine pronouns outnumber feminine ones by roughly 3:1.
    • A Balanced BUG subset contains 25,844 sentences.
  • Filtering and labelling: matches were kept only if they contained one of 183 census professions. Gendered nouns (e.g., "sister") and non-human nouns were removed. Labels follow the 2015 U.S. census gender distribution per occupation.
  • Quality checks: 17 annotators reviewed 1,700 sampled sentences and judged 85% correct. These annotations are released as Gold BUG. Inter-annotator agreement was 95.5% (κ = 0.73).
  • Translation findings:
    • Average translation accuracy on BUG was 69.9%, compared with 47.6% on earlier synthetic sets. However, the relative bias gaps (∆G, ∆S) were similar to those reported before.
    • Czech was the easiest target language. Russian showed the largest ∆G and Hebrew the largest ∆S.
    • Bias grew as the number of pronouns in a sentence increased.
  • Coreference findings: SpanBERT's ∆S was 6.0, compared with 13.5 for older models. Fine-tuning on anti-stereotypical data cut errors by more than 50% (3 points absolute) at a cost of about 1 point of overall accuracy. Data and models are available at github.com/SLAB-NLP/BUG.

Our summary of the paper, in our own words — not its text. The work belongs to its authors; read it at the links above.

Fields: Linguistics & languages

Topics: Coreference resolution · Machine translation

Cited by

46 papers of our library cite it (of the 140673 whose bibliographies we read); 6 of them have a page here, 40 not yet.

Cite

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.18653/v1/2021.findings-emnlp.211 · arXiv 2109.03858

@inproceedings{levy2021collecting,
  title = {{Collecting a Large-Scale Gender Bias Dataset for Coreference Resolution and Machine Translation}},
  author = {Levy, Shahar and Lazar, Koren and Stanovsky, Gabriel},
  booktitle = {Findings of the Association for Computational Linguistics: EMNLP 2021},
  year = {2021},
  doi = {10.18653/v1/2021.findings-emnlp.211},
  eprint = {2109.03858},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.18653/v1/2021.findings-emnlp.211}
}

References

24 references, 12 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Mandar Joshi, Danqi Chen, Yinhan Liu, Daniel S. Weld, Luke Zettlemoyer, and Omer Levy. 2020. SpanBERT: Improving pre-training by representing and predicting spans. Transactions of the Association for Computational Linguistics, 8:64–77. find our card
  2. Emily M. Bender, Timnit Gebru, Angelina McMillan-Major, and Shmargaret Shmitchell. 2021. On the dangers of stochastic parrots: Can language models be too big? . In Proceedings of the 2021 ACM Conference on Fairness, Accountability, and Trans-Mandar Joshi, Omer Levy, Luke Zettlemoyer, and Daniel Weld… find our card
  3. Svetlana Kiritchenko and Saif Mohammad. 2018. Examining gender and race bias in two hundred sentiment analysis systems. In Proceedings of the Seventh Joint Conference on Lexical and Computational Semantics, pages 43–53, New Orleans, Louisiana. Association for Computational Linguistics. find our card
  4. Vid Kocijan, Oana-Maria Camburu, and Thomas Lukasiewicz. 2021. The gap on gap: Tackling the problem of differing data distributions in bias-measuring datasets. In AAAI. find
  5. Tom Kocmi, Tomasz Limisiewicz, and Gabriel Stanovsky. 2020. Gender coreference and bias evaluation at WMT 2020. In Proceedings of the Fifth Conference on Machine Translation, pages 357–364, Online. Association for Computational Linguistics. find
  6. Kenton Lee, Luheng He, and Luke Zettlemoyer. 2018. Higher-order coreference resolution with coarse-to-fine inference. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers), pages 687–… find our card
  7. Hector Levesque, Ernest Davis, and Leora Morgenstern. 2012. The winograd schema challenge. In Thirteenth International Conference on the Principles of Knowledge Representation and Reasoning. Citeseer. find
  8. Yinhan Liu, Jiatao Gu, Naman Goyal, Xian Li, Sergey Edunov, Marjan Ghazvininejad, Mike Lewis, and Luke Zettlemoyer. 2020. Multilingual denoising pre-training for neural machine translation. Transactions of the Association for Computational Linguistics, 8:726–742. find
  9. Ninareh Mehrabi, Fred Morstatter, Nripsuta Saxena, Kristina Lerman, and Aram Galstyan. 2021. A survey on bias and fairness in machine learning. ACM Comput. Surv., 54(6). find our card
  10. Alexandra Olteanu, Carlos Castillo, Fernando Diaz, and Emre Kıcıman. 2019. Social data: Biases, methodological pitfalls, and ethical boundaries. Frontiers in Big Data, 2:13. find
  11. Sameer Pradhan, Alessandro Moschitti, Nianwen Xue, Olga Uryupina, and Yuchen Zhang. 2012. CoNLL-2012 shared task: Modeling multilingual unrestricted coreference in OntoNotes. In Joint Conference on EMNLP and CoNLL - Shared Task, pages 1–40, Jeju Island, Korea. Association for Computational Linguisti… find our card
  12. Marcelo O. R. Prates, Pedro H. C. Avelar, and L. Lamb. 2019. Assessing gender bias in machine translation: a case study with google translate. Neural Computing and Applications, 32:6363–6381. find
  13. Adithya Renduchintala and Adina Williams. 2021. Investigating failures of automatic translation in the case of unambiguous gender. ArXiv, abs/2104.07838. find
  14. Rachel Rudinger, Chandler May, and Benjamin Van Durme. 2017. Social bias in elicited natural language inferences. In Proceedings of the First ACL Workshop on Ethics in Natural Language Processing, pages 74–79, Valencia, Spain. Association for Computational Linguistics. find our card
  15. Rachel Rudinger, Jason Naradowsky, Brian Leonard, and Benjamin Van Durme. 2018. Gender bias in coreference resolution. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 2 (Short Papers), pages 8–1… find our card
  16. Danielle Saunders, Rosie Sallis, and Bill Byrne. 2020. Neural machine translation doesn’t translate gender coreference right unless you make it. In Proceedings of the Second Workshop on Gender Bias in Natural Language Processing, pages 35–43, Barcelona, Spain (Online). Association for Computational… find our card
  17. Micah Shlain, Hillel Taub-Tabib, Shoval Sadde, and Yoav Goldberg. 2020. Syntactic search by example. In Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics: System Demonstrations, pages 17–23, Online. Association for Computational Linguistics. find
  18. Gabriel Stanovsky, Noah A. Smith, and Luke Zettlemoyer. 2019. Evaluating gender bias in machine translation. In Proceedings of the 57th Annual Meeting of the Association for Computational Linguistics, pages 1679–1684, Florence, Italy. Association for Computational Linguistics. find our card
  19. Harini Suresh and John V. Guttag. 2021. A framework for understanding sources of harm throughout the machine learning life cycle. find
  20. Yuqing Tang, Chau Tran, Xian Li, Peng-Jen Chen, Naman Goyal, Vishrav Chaudhary, Jiatao Gu, and Angela Fan. 2020. Multilingual translation with extensible multilingual pretraining and finetuning. find
  21. Jörg Tiedemann and Santhosh Thottingal. 2020. OPUS-MT — Building open translation services for the World. In Proceedings of the 22nd Annual Conferenec of the European Association for Machine Translation (EAMT), Lisbon, Portugal. find
  22. Lucy Lu Wang, Kyle Lo, Yoganand Chandrasekhar, Russell Reas, Jiangjiang Yang, Doug Burdick, Darrin Eide, Kathryn Funk, Yannis Katsis, Rodney Michael Kinney, Yunyao Li, Ziyang Liu, William Merrill, Paul Mooney, Dewey A. Murdick, Devvret Rishi, Jerry Sheehan, Zhihong Shen, Brandon Stilson, Alex D. Wad… find
  23. Kellie Webster, Marta Recasens, Vera Axelrod, and Jason Baldridge. 2018. Mind the GAP: A balanced corpus of gendered ambiguous pronouns. Transactions of the Association for Computational Linguistics, 6:605–617. find our card
  24. Jieyu Zhao, Tianlu Wang, Mark Yatskar, Vicente Ordonez, and Kai-Wei Chang. 2018. Gender bias in coreference resolution: Evaluation and debiasing methods. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies,… find our card

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/levy-2021-collecting-large-scale-gender-bias/