Detecting Word Sense Disambiguation Biases in Machine Translation for Model-Agnostic Adversarial Attacks

Denis Emelin, Ivan Titov, Rico Sennrich


Abstract
Word sense disambiguation is a well-known source of translation errors in NMT. We posit that some of the incorrect disambiguation choices are due to models’ over-reliance on dataset artifacts found in training data, specifically superficial word co-occurrences, rather than a deeper understanding of the source text. We introduce a method for the prediction of disambiguation errors based on statistical data properties, demonstrating its effectiveness across several domains and model types. Moreover, we develop a simple adversarial attack strategy that minimally perturbs sentences in order to elicit disambiguation errors to further probe the robustness of translation models. Our findings indicate that disambiguation robustness varies substantially between domains and that different models trained on the same data are vulnerable to different attacks.
Anthology ID:
2020.emnlp-main.616
Volume:
Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing (EMNLP)
Month:
November
Year:
2020
Address:
Online
Venue:
EMNLP
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
7635–7653
Language:
URL:
https://aclanthology.org/2020.emnlp-main.616
DOI:
10.18653/v1/2020.emnlp-main.616
Bibkey:
Copy Citation:
PDF:
https://aclanthology.org/2020.emnlp-main.616.pdf
Video:
 https://slideslive.com/38939052
Code
 demelin/detecting_wsd_biases_for_nmt