Towards Robust NLG Bias Evaluation with Syntactically-diverse Prompts

Arshiya Aggarwal, Jiao Sun, Nanyun Peng


Abstract
We present a robust methodology for evaluating biases in natural language generation(NLG) systems. Previous works use fixed hand-crafted prefix templates with mentions of various demographic groups to prompt models to generate continuations for bias analysis. These fixed prefix templates could themselves be specific in terms of styles or linguistic structures, which may lead to unreliable fairness conclusions that are not representative of the general trends from tone varying prompts. To study this problem, we paraphrase the prompts with different syntactic structures and use these to evaluate demographic bias in NLG systems. Our results suggest similar overall bias trends but some syntactic structures lead to contradictory conclusions compared to past works. We show that our methodology is more robust and that some syntactic structures prompt more toxic content while others could prompt less biased generation. This suggests the importance of not relying on a fixed syntactic structure and using tone-invariant prompts. Introducing syntactically-diverse prompts can achieve more robust NLG (bias) evaluation.
Anthology ID:
2022.findings-emnlp.445
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:
6022–6032
Language:
URL:
https://aclanthology.org/2022.findings-emnlp.445
DOI:
10.18653/v1/2022.findings-emnlp.445
Bibkey:
Cite (ACL):
Arshiya Aggarwal, Jiao Sun, and Nanyun Peng. 2022. Towards Robust NLG Bias Evaluation with Syntactically-diverse Prompts. In Findings of the Association for Computational Linguistics: EMNLP 2022, pages 6022–6032, Abu Dhabi, United Arab Emirates. Association for Computational Linguistics.
Cite (Informal):
Towards Robust NLG Bias Evaluation with Syntactically-diverse Prompts (Aggarwal et al., Findings 2022)
Copy Citation:
PDF:
https://aclanthology.org/2022.findings-emnlp.445.pdf
Video:
 https://aclanthology.org/2022.findings-emnlp.445.mp4