Rare but Severe Neural Machine Translation Errors Induced by Minimal Deletion: An Empirical Study on Chinese and English
This work addresses robustness issues in neural machine translation for Chinese and English, but it is incremental as it focuses on empirical analysis of existing models.
The study investigated how minimal deletions (single characters or words) in source text induce rare but severe errors in Chinese-English neural machine translation, finding that word deletions hurt overall scores more, but character deletions cause specific errors, with effects varying by language direction and training data size.
We examine the inducement of rare but severe errors in English-Chinese and Chinese-English in-domain neural machine translation by minimal deletion of the source text with character-based models. By deleting a single character, we can induce severe translation errors. We categorize these errors and compare the results of deleting single characters and single words. We also examine the effect of training data size on the number and types of pathological cases induced by these minimal perturbations, finding significant variation. We find that deleting a word hurts overall translation score more than deleting a character, but certain errors are more likely to occur when deleting characters, with language direction also influencing the effect.