MedNLI Is Not Immune: Natural Language Inference Artifacts in the Clinical Domain

Christine Herlihy, Rachel Rudinger


Abstract
Crowdworker-constructed natural language inference (NLI) datasets have been found to contain statistical artifacts associated with the annotation process that allow hypothesis-only classifiers to achieve better-than-random performance (CITATION). We investigate whether MedNLI, a physician-annotated dataset with premises extracted from clinical notes, contains such artifacts (CITATION). We find that entailed hypotheses contain generic versions of specific concepts in the premise, as well as modifiers related to responsiveness, duration, and probability. Neutral hypotheses feature conditions and behaviors that co-occur with, or cause, the condition(s) in the premise. Contradiction hypotheses feature explicit negation of the premise and implicit negation via assertion of good health. Adversarial filtering demonstrates that performance degrades when evaluated on the difficult subset. We provide partition information and recommendations for alternative dataset construction strategies for knowledge-intensive domains.
Anthology ID:
2021.acl-short.129
Volume:
Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers)
Month:
August
Year:
2021
Address:
Online
Editors:
Chengqing Zong, Fei Xia, Wenjie Li, Roberto Navigli
Venues:
ACL | IJCNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
1020–1027
Language:
URL:
https://aclanthology.org/2021.acl-short.129
DOI:
10.18653/v1/2021.acl-short.129
Bibkey:
Cite (ACL):
Christine Herlihy and Rachel Rudinger. 2021. MedNLI Is Not Immune: Natural Language Inference Artifacts in the Clinical Domain. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers), pages 1020–1027, Online. Association for Computational Linguistics.
Cite (Informal):
MedNLI Is Not Immune: Natural Language Inference Artifacts in the Clinical Domain (Herlihy & Rudinger, ACL-IJCNLP 2021)
Copy Citation:
PDF:
https://aclanthology.org/2021.acl-short.129.pdf
Optional supplementary material:
 2021.acl-short.129.OptionalSupplementaryMaterial.zip
Video:
 https://aclanthology.org/2021.acl-short.129.mp4
Code
 crherlihy/clinical_nli_artifacts
Data
MIMIC-IIIMultiNLISNLI