National Research Council of Canada. Information and Communication Technologies
The 53rd Annual Meeting of the Association for Computational Linguistics and the 7the International Joint Conference on Natural Language Processing, July 26-31, 2015, Beijing, China
Precisely evaluating the quality of a translation against human references is a challenging task due to the flexible word ordering of a sentence and the existence of a large number of synonyms for words. This paper proposes to evaluate translations with distributed representations of words and sentences. We study several metrics based on word and sentence representations and their combination. Experiments on the WMT metric task shows that the metric based on the combined representations achieves the best performance, outperforming the state-of-the-art translation metrics by a large margin. In particular , training the distributed representations only needs a reasonable amount of mono-lingual, unlabeled data that is not necessary drawn from the test domain.
Association for Computational Linguistics
Proceedings of the 53rd Annual Meeting of the Association for Computational Linguistics and the 7the International Joint Conference on Natural Language Processing (Short Papers), Beijing, China, July 26-31, 2015: 150–155.