Abstract
Approaches to Quality Estimation (QE) of machine translation have shown promising results at predicting quality scores for translated sentences. However, QE models are often trained on noisy approximations of quality annotations derived from the proportion of post-edited words in translated sentences instead of direct human annotations of translation errors. The latter is a more reliable ground-truth but more expensive to obtain. In this paper, we present the first attempt to model the task of predicting the proportion of actual translation errors in a sentence while minimising the need for direct human annotation. For that purpose, we use transfer-learning to leverage large scale noisy annotations and small sets of high-fidelity human annotated translation errors to train QE models. Experiments on four language pairs and translations obtained by statistical and neural models show consistent gains over strong baselines.Citation
Blain, F., Aletras, N. and Specia, L. (2020) Quality in, quality out: learning from actual mistakes. In Proceedings of the 22nd Annual Conference of the European Association for Machine Translation (EAMT), edited by Forcada, M. L., Martins, A., Moniz, H., Turchi, M. et al.,Lisbon, Portugal: European Association for Machine Translation.Additional Links
https://www.aclweb.org/anthology/2020.eamt-1.16/Type
Conference contributionLanguage
enDescription
© 2020 The Authors. Published by European Association for Machine Translation. This is an open access article available under a Creative Commons licence. The published version can be accessed at the following link on the publisher’s website: https://www.aclweb.org/anthology/2020.eamt-1.16/Sponsors
This work was supported by the Bergamot project (EU H2020 Grant No. 825303).
Except where otherwise noted, this item's license is described as https://creativecommons.org/licenses/by-nd/3.0/