BLEU might be Guilty but References are not Innocent
This work addresses a critical issue for researchers and practitioners in machine translation by improving evaluation reliability, though it is incremental as it builds on existing metric limitations.
The paper tackles the problem of poor correlation between automatic metrics and human judgment in machine translation evaluation, showing that diversifying reference translations through paraphrasing improves correlation with human evaluation across various systems and metrics, including for high-quality outputs where standard references fail.
The quality of automatic metrics for machine translation has been increasingly called into question, especially for high-quality systems. This paper demonstrates that, while choice of metric is important, the nature of the references is also critical. We study different methods to collect references and compare their value in automated evaluation by reporting correlation with human evaluation for a variety of systems and metrics. Motivated by the finding that typical references exhibit poor diversity, concentrating around translationese language, we develop a paraphrasing task for linguists to perform on existing reference translations, which counteracts this bias. Our method yields higher correlation with human judgment not only for the submissions of WMT 2019 English to German, but also for Back-translation and APE augmented MT output, which have been shown to have low correlation with automatic metrics using standard references. We demonstrate that our methodology improves correlation with all modern evaluation metrics we look at, including embedding-based methods. To complete this picture, we reveal that multi-reference BLEU does not improve the correlation for high quality output, and present an alternative multi-reference formulation that is more effective.