Elamur Research⌕
☰
home / papers / 2023 / Interpretable Unified Language Checking

Interpretable Unified Language Checking

Paper · 2023 · Tianhua Zhang, Hongyin Luo, Yung-Sung Chuang et al.

Summary

Harmful text, whether written by people or produced by LLMs, can be wrong in different ways: it may be false, stereotyping, or hateful. Earlier work usually built a separate supervised model for each of these problems, so the system had to know the task in advance. The authors propose UniLC, a task-agnostic checker that handles all three problems through one shared few-shot prompt. GPT3.5-turbo works in two steps. First, it summarizes what looks problematic in a claim and generates a grounding fact, such as a scientific or social fact. Second, it decides whether the claim agrees with that grounding. The second step can be done by the LLM itself or by a separate entailment classifier. The authors compare zero-shot checking with few-shot fact generation combined with either zero-shot or few-shot classification. They evaluate on six fact-checking, stereotype and hate-speech datasets. The unified prompt matches or beats fully supervised baselines on several of them, and the generated groundings make each verdict explainable.

Key points

  • The paper tests three prompting setups: a zero-shot baseline, few-shot fact prediction plus zero-shot classification, and few-shot fact prediction plus few-shot classification.
  • A single prompt mixes exemplars from fact-checking and fairness tasks without naming the task. The extracted text calls this setup "12-shot," which is probably a garbled fraction.
  • The datasets include Climate-fever, PubHealth, ToxiGen (machine-generated hate speech) and MGFN. For MGFN, groundings are extracted from the provided documents instead of being generated.
  • Using a non-LLM entailment classifier for the final stance step consistently improved results, especially on the fairness tasks.
  • Grounding categories overlap across tasks: "social" facts were the most common everywhere except Climate-fever. Code and data are available at github.com/luohongyin/UniLC.

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

 Institutions

🎓Massachusetts Institute of TechnologyCambridge · United States211 works · 69,000 citations🔬MIT Computer Science and Artificial Intelligence LaboratoryCambridge · United States69 works · 30,953 citations🎓Chinese University of Hong KongHong Kong · China126 works · 14,073 citations🔬Centre for Perceptual and Interactive IntelligenceHong Kong · China8 works · 1,209 citations

 Fields, topics and concepts

📚 Shelves 6The field of science this work belongs to (OpenAlex fields).Computer Science · 16,667 worksArtificial Intelligence · 16,194 worksArts and Humanities · 16,039 worksLanguage and Linguistics · 15,891 worksSocial Sciences · 1,772 worksSociology and Political Science · 577 works

🧭 Topics 1Research subjects and languages the work is about.Bias, fairness & safety · 517 works

Cite

Cite as Tianhua Zhang, Hongyin Luo, Yung-Sung Chuang, Wei Fang, Luc Gaitskell, Thomas Hartvigsen, Xixin Wu, Danny Fox, Helen Meng and James Glass. 2023. Interpretable Unified Language Checking. arXiv:2304.03728.

The paper itself, not this page. BibTeX from the library's record of the work. DOI 10.48550/arxiv.2304.03728 · arXiv 2304.03728

@misc{zhang2023interpretable,
  title = {{Interpretable Unified Language Checking}},
  author = {Tianhua Zhang, Hongyin Luo, Yung-Sung Chuang, Wei Fang, Luc Gaitskell, Thomas Hartvigsen, Xixin Wu, Danny Fox, Helen Meng and James Glass},
  year = {2023},
  doi = {10.48550/arxiv.2304.03728},
  eprint = {2304.03728},
  archivePrefix = {arXiv},
  url = {https://doi.org/10.48550/arxiv.2304.03728}
}

References

44 references, 28 of them with a card of ours. ↗ leads to the work, «find» searches for one we have no id for yet.

  1. Abubakar Abid, Maheen Farooqi, and James Zou. 2021. Large language models associate muslims with violence. Nature Machine Intelligence, 3(6):461–463. find
  2. Rami Aly, Zhijiang Guo, Michael Schlichtkrull, James Thorne, Andreas Vlachos, Christos Christodoulopoulos, Oana Cocarascu, and Arpit Mittal. 2021. Feverous: Fact extraction and verification over unstructured and structured information. In 35th Conference on Neural Information Processing Systems, Neu… find
  3. Amanda Askell, Yuntao Bai, Anna Chen, Dawn Drain, Deep Ganguli, Tom Henighan, Andy Jones, Nicholas Joseph, Ben Mann, Nova DasSarma, et al. 2021. A general language assistant as a laboratory for alignment. arXiv preprint arXiv:2112.00861. find
  4. Isabelle Augenstein, Tim Rocktäschel, Andreas Vlachos, and Kalina Bontcheva. 2016. Stance detection with bidirectional conditional encoding. In Proceedings of the 2016 Conference on Empirical Methods in Natural Language Processing, pages 876–885.
  5. Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901.
  6. Ke-Li Chiu, Annie Collins, and Rohan Alexander. 2021a. Detecting hate speech with gpt-3. arXiv preprint arXiv:2103.12407. — [arXiv:2103.12407](https://arxiv.org/abs/2103.12407) find
  7. Thomas KF Chiu, Helen Meng, Ching-Sing Chai, Irwin King, Savio Wong, and Yeung Yam. 2021b. Creation and evaluation of a pretertiary artificial intelligence (ai) curriculum. IEEE Transactions on Education, 65(1):30–39. find
  8. Aakanksha Chowdhery, Sharan Narang, Jacob Devlin, Maarten Bosma, Gaurav Mishra, Adam Roberts, Paul Barham, Hyung Won Chung, Charles Sutton, Sebastian Gehrmann, et al. 2022. Palm: Scaling language modeling with pathways. arXiv preprint arXiv:2204.02311.
  9. Antonia Creswell and Murray Shanahan. 2022. Faithful reasoning using large language models. arXiv preprint arXiv:2208.14271. find
  10. Ona de Gibert, Naiara Perez, Aitor Garcıa-Pablos, and Montse Cuadros. 2018a. Hate speech dataset from a white supremacy forum.
  11. Ona de Gibert, Naiara Perez, Aitor García-Pablos, and Montse Cuadros. 2018b. Hate Speech Dataset from a White Supremacy Forum. In Proceedings of the 2nd Workshop on Abusive Language Online (ALW2), pages 11–20, Brussels, Belgium. Association for Computational Linguistics.
  12. Thomas Diggelmann, Jordan Boyd-Graber, Jannis Bulian, Massimiliano Ciaramita, and Markus Leippold. 2020. Climate-fever: A dataset for verification of real-world climate claims. arXiv preprint arXiv:2012.00614.
  13. Nemanja Djuric, Jing Zhou, Robin Morris, Mihajlo Gr-bovic, Vladan Radosavljevic, and Narayan Bhamidipati. 2015. Hate speech detection with comment embeddings. In Proceedings of the 24th international conference on world wide web, pages 29–30. find
  14. Deep Ganguli, Amanda Askell, Nicholas Schiefer, Thomas Liao, Kamilė Lukošiūtė, Anna Chen, Anna Goldie, Azalia Mirhoseini, Catherine Olsson, Danny Hernandez, et al. 2023. The capacity for moral self-correction in large language models. arXiv preprint arXiv:2302.07459.
  15. Deep Ganguli, Danny Hernandez, Liane Lovitt, Amanda Askell, Yuntao Bai, Anna Chen, Tom Conerly, Nova Dassarma, Dawn Drain, Nelson Elhage, et al. 2022. Predictability and surprise in large generative models. In 2022 ACM Conference on Fairness, Accountability, and Transparency, pages 1747– 1764. find
  16. Akash Gautam, Puneet Mathur, Rakesh Gosangi, Debanjan Mahata, Ramit Sawhney, and Rajiv Ratn Shah. 2020. \# metooma: Multi-aspect annotations of tweets related to the metoo movement. In Proceedings of the International AAAI Conference on Web and Social Media, volume 14, pages 209–216. find
  17. Samuel Gehman, Suchin Gururangan, Maarten Sap, Yejin Choi, and Noah A Smith. 2020. Realtoxicityprompts: Evaluating neural toxic degeneration in language models. In Findings of the Association for Computational Linguistics: EMNLP 2020, pages 3356–3369.
  18. Zhijiang Guo, Michael Sejr Schlichtkrull, and Andreas Vlachos. 2022. A survey on automated fact-checking. Transactions of the Association for Computational Linguistics, 10:178–206.
  19. Thomas Hartvigsen, Saadia Gabriel, Hamid Palangi, Maarten Sap, Dipankar Ray, and Ece Kamar. 2022. Toxigen: A large-scale machine-generated dataset for adversarial and implicit hate speech detection. In Proceedings of the 60th Annual Meeting of the Association for Computational Linguistics (Volume 1:…
  20. Ziwei Ji, Nayeon Lee, Rita Frieske, Tiezheng Yu, Dan Su, Yan Xu, Etsuko Ishii, Yejin Bang, Andrea Madotto, and Pascale Fung. 2022. Survey of hallucination in natural language generation. ACM Computing Surveys.
  21. Liwei Jiang, Jena D Hwang, Chandra Bhagavatula, Ronan Le Bras, Maxwell Forbes, Jon Borchardt, Jenny Liang, Oren Etzioni, Maarten Sap, and Yejin Choi. 2021. Delphi: Towards machine ethics and norms. arXiv preprint arXiv:2110.07574. find
  22. Neema Kotonya and Francesca Toni. 2020. Explainable automated fact-checking for public health claims. In Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP), pages 7740–7754, Online. Association for Computational Linguistics.
  23. Kaiji Lu, Piotr Mardziel, Fangjing Wu, Preetam Amancharla, and Anupam Datta. 2020. Gender bias in neural natural language processing. Logic, Language, and Security: Essays Dedicated to Andre Scedrov on the Occasion of His 65th Birthday, pages 189–202.
  24. Hongyin Luo and James Glass. 2023. Logic against bias: Textual entailment mitigates stereotypical sentence reasoning. arXiv preprint arXiv:2303.05670. find
  25. Sean MacAvaney, Hao-Ren Yao, Eugene Yang, Katina Russell, Nazli Goharian, and Ophir Frieder. 2019. Hate speech detection: Challenges and solutions. PloS one, 14(8):e0221152. find
  26. Potsawee Manakul, Adian Liusie, and Mark JF Gales. 2023. Selfcheckgpt: Zero-resource black-box hallucination detection for generative large language models. arXiv preprint arXiv:2303.08896. find
  27. Moin Nadeem, Anna Bethke, and Siva Reddy. 2021. Stereoset: Measuring stereotypical bias in pre-trained language models. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Lon… find
  28. Moin Nadeem, Wei Fang, Brian Xu, Mitra Mohtarami, and James Glass. 2019. Fakta: An automatic end-to-end fact checking system. arXiv preprint arXiv:1906.04164. find
  29. Long Ouyang, Jeff Wu, Xu Jiang, Diogo Almeida, Carroll L Wainwright, Pamela Mishkin, Chong Zhang, Sandhini Agarwal, Katarina Slama, Alex Ray, et al. 2022. Training language models to follow instructions with human feedback. arXiv preprint arXiv:2203.02155.
  30. Marta Recasens, Cristian Danescu-Niculescu-Mizil, and Dan Jurafsky. 2013. Linguistic models for analyzing and detecting biased language. In Proceedings of the 51st Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers), pages 1650–1659.
  31. Paul Röttger, Bertram Vidgen, Dong Nguyen, Zeerak Waseem, Helen Margetts, and Janet B Pierrehumbert. 2020. Hatecheck: Functional tests for hate speech detection models. arXiv preprint arXiv:2012.15606.
  32. Maarten Sap, Saadia Gabriel, Lianhui Qin, Dan Jurafsky, Noah A Smith, and Yejin Choi. 2020. Social bias frames: Reasoning about social and power implications of language. In ACL.
  33. Tal Schuster, Roei Schuster, Darsh J. Shah, and Regina Barzilay. 2020. The limitations of stylometry for detecting machine-generated fake news. Computational Linguistics, 46(2):499–510.
  34. Kurt Shuster, Spencer Poff, Moya Chen, Douwe Kiela, and Jason Weston. 2021. Retrieval augmentation reduces hallucination in conversation. In Findings of the Association for Computational Linguistics: EMNLP 2021, pages 3784–3803.
  35. Romal Thoppilan, Daniel De Freitas, Jamie Hall, Noam Shazeer, Apoorv Kulshreshtha, Heng-Tze Cheng, Alicia Jin, Taylor Bos, Leslie Baker, Yu Du, et al. 2022. Lamda: Language models for dialog applications. arXiv preprint arXiv:2201.08239.
  36. Adam Trischler, Tong Wang, Xingdi Yuan, Justin Harris, Alessandro Sordoni, Philip Bachman, and Kaheer Suleman. 2017. NewsQA: A machine comprehension dataset. In Proceedings of the 2nd Workshop on Representation Learning for NLP, pages 191–200, Vancouver, Canada. Association for Computational Linguis…
  37. William Yang Wang. 2017. “liar, liar pants on fire”: A new benchmark dataset for fake news detection. In Proceedings of the 55th Annual Meeting of the Association for Computational Linguistics (Volume 2: Short Papers), pages 422–426.
  38. Xuezhi Wang, Jason Wei, Dale Schuurmans, Quoc Le, Ed Chi, and Denny Zhou. 2022. Self-consistency improves chain of thought reasoning in language models. arXiv preprint arXiv:2203.11171.
  39. Kellie Webster, Xuezhi Wang, Ian Tenney, Alex Beutel, Emily Pitler, Ellie Pavlick, Jilin Chen, Ed Chi, and Slav Petrov. 2020. Measuring and reducing gendered correlations in pre-trained models. arXiv preprint arXiv:2010.06032.
  40. Jason Wei, Maarten Bosma, Vincent Y Zhao, Kelvin Guu, Adams Wei Yu, Brian Lester, Nan Du, Andrew M Dai, and Quoc V Le. 2021. Finetuned language models are zero-shot learners. arXiv preprint arXiv:2109.01652.
  41. Jason Wei, Xuezhi Wang, Dale Schuurmans, Maarten Bosma, Fei Xia, Ed H Chi, Quoc V Le, Denny Zhou, et al. 2022. Chain-of-thought prompting elicits reasoning in large language models. In Advances in Neural Information Processing Systems.
  42. Adina Williams, Nikita Nangia, and Samuel Bowman. 2018. A broad-coverage challenge corpus for sentence understanding through inference. In Proceedings of the 2018 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, Volume 1 (Long Pa…
  43. Wenjie Yin and Arkaitz Zubiaga. 2021. Towards generalisable hate speech detection: a review on obstacles and solutions. PeerJ Computer Science, 7:e598. find
  44. Rowan Zellers, Ari Holtzman, Hannah Rashkin, Yonatan Bisk, Ali Farhadi, Franziska Roesner, and Yejin Choi. 2020. Defending against neural fake news.

Updated

↑ back to top

This page in Markdown → https://md.research.elamur.ai/free/en/papers/zhang-2023-interpretable-unified-language-checking/