Abstract
We investigate MT evaluation metric performance on adversarially-synthesized texts, to shed light on metric robustness. We experiment with word- and character-level attacks on three popular machine translation metrics: BERTScore, BLEURT, and COMET. Our human experiments validate that automatic metrics tend to overpenalize adversarially-degraded translations. We also identify inconsistencies in BERTScore ratings, where it judges the original sentence and the adversarially-degraded one as similar, while judging the degraded translation as notably worse than the original with respect to the reference. We identify patterns of brittleness that motivate more robust metric development.- Anthology ID:
- 2023.findings-emnlp.340
- Volume:
- Findings of the Association for Computational Linguistics: EMNLP 2023
- Month:
- December
- Year:
- 2023
- Address:
- Singapore
- Editors:
- Houda Bouamor, Juan Pino, Kalika Bali
- Venue:
- Findings
- SIG:
- Publisher:
- Association for Computational Linguistics
- Note:
- Pages:
- 5126–5135
- Language:
- URL:
- https://aclanthology.org/2023.findings-emnlp.340
- DOI:
- 10.18653/v1/2023.findings-emnlp.340
- Cite (ACL):
- Yichen Huang and Timothy Baldwin. 2023. Robustness Tests for Automatic Machine Translation Metrics with Adversarial Attacks. In Findings of the Association for Computational Linguistics: EMNLP 2023, pages 5126–5135, Singapore. Association for Computational Linguistics.
- Cite (Informal):
- Robustness Tests for Automatic Machine Translation Metrics with Adversarial Attacks (Huang & Baldwin, Findings 2023)
- PDF:
- https://preview.aclanthology.org/emnlp-22-attachments/2023.findings-emnlp.340.pdf