Abstract
I highlight a simple failure mode of state-of-the-art machine reading systems: when contexts do not align with commonly shared beliefs. For example, machine reading systems fail to answer What did Elizabeth want? correctly in the context of ‘My kingdom for a cough drop, cried Queen Elizabeth.’ Biased by co-occurrence statistics in the training data of pretrained language models, systems predict my kingdom, rather than a cough drop. I argue such biases are analogous to human belief biases and present a carefully designed challenge dataset for English machine reading, called Auto-Locke, to quantify such effects. Evaluations of machine reading systems on Auto-Locke show the pervasiveness of belief bias in machine reading.- Anthology ID:
- 2021.emnlp-main.649
- Volume:
- Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing
- Month:
- November
- Year:
- 2021
- Address:
- Online and Punta Cana, Dominican Republic
- Venue:
- EMNLP
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 8240–8245
- Language:
- URL:
- https://aclanthology.org/2021.emnlp-main.649
- DOI:
- 10.18653/v1/2021.emnlp-main.649
- Cite (ACL):
- Anders Søgaard. 2021. Locke’s Holiday: Belief Bias in Machine Reading. In Proceedings of the 2021 Conference on Empirical Methods in Natural Language Processing, pages 8240–8245, Online and Punta Cana, Dominican Republic. Association for Computational Linguistics.
- Cite (Informal):
- Locke’s Holiday: Belief Bias in Machine Reading (Søgaard, EMNLP 2021)
- PDF:
- https://preview.aclanthology.org/remove-xml-comments/2021.emnlp-main.649.pdf
- Data
- DROP