Tagged Back-translation Revisited: Why Does It Really Work?
Tagged Back-translation Revisited: Why Does It Really Work?
复制标题
DOI：
10.18653/v1/2020.acl-main.532
复制
发表时间：
2020-07
影响因子：
--
通讯作者：
Benjamin Marie;Raphaël Rubino;Atsushi Fujita
中科院分区：
文献类型：
--
作者：
Benjamin Marie;Raphaël Rubino;Atsushi Fujita
在本文中，我们展示了在大型反向翻译数据上训练的神经机器翻译（NMT）系统过度拟合了机器翻译文本的一些特征。这种 NMT 系统可以更好地翻译人工翻译，即翻译语，但可能会大大降低原始文本的翻译质量。我们的分析表明，在反向翻译中添加一个简单的标签可以防止这种质量下降，并通过帮助 NMT 系统在训练期间区分反向翻译数据和原始并行数据，平均提高整体翻译质量。我们还表明，与高资源配置相比，在低资源环境中训练的 NMT 系统更不容易受到过度拟合反向翻译的影响。我们的结论是，训练数据中的反向翻译应该始终被标记，特别是当要翻译的文本的来源未知时。
In this paper, we show that neural machine translation (NMT) systems trained on large back-translated data overfit some of the characteristics of machine-translated texts. Such NMT systems better translate human-produced translations, i.e., translationese, but may largely worsen the translation quality of original texts. Our analysis reveals that adding a simple tag to back-translations prevents this quality degradation and improves on average the overall translation quality by helping the NMT system to distinguish back-translated data from original parallel data during training. We also show that, in contrast to high-resource configurations, NMT systems trained in low-resource settings are much less vulnerable to overfit back-translations. We conclude that the back-translations in the training data should always be tagged especially when the origin of the text to be translated is unknown.