Contextualizing Language Models for Norms Diverging from Social Majority

Niklas Kiehne, Hermann Kroll, Wolf-Tilo Balke


Abstract
To comprehensibly contextualize decisions, artificial systems in social situations need a high degree of awareness of the rules of conduct of human behavior. Especially transformer-based language models have recently been shown to exhibit some such awareness. But what if norms in some social setting do not adhere to or even blatantly deviate from the mainstream? In this paper, we introduce a novel mechanism based on deontic logic to allow for a flexible adaptation of individual norms by de-biasing training data sets and a task-reduction to textual entailment. Building on the popular ‘Moral Stories’ dataset we on the one hand highlight the intrinsic bias of current language models, on the other hand characterize the adaptability of pre-trained models to deviating norms in fine-tuning settings.
Anthology ID:
2022.findings-emnlp.339
Volume:
Findings of the Association for Computational Linguistics: EMNLP 2022
Month:
December
Year:
2022
Address:
Abu Dhabi, United Arab Emirates
Editors:
Yoav Goldberg, Zornitsa Kozareva, Yue Zhang
Venue:
Findings
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
4620–4633
Language:
URL:
https://aclanthology.org/2022.findings-emnlp.339
DOI:
10.18653/v1/2022.findings-emnlp.339
Bibkey:
Cite (ACL):
Niklas Kiehne, Hermann Kroll, and Wolf-Tilo Balke. 2022. Contextualizing Language Models for Norms Diverging from Social Majority. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 4620–4633, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
Contextualizing Language Models for Norms Diverging from Social Majority (Kiehne et al., Findings 2022)
Copy Citation:
PDF:
https://preview.aclanthology.org/nschneid-patch-4/2022.findings-emnlp.339.pdf
Dataset:
 2022.findings-emnlp.339.dataset.zip
Software:
 2022.findings-emnlp.339.software.zip