Selective Memory-Augmented Document Translation with Diverse Global Context
Xu Zhang, Jinlong Li, Mingzhi Liu · 2022
Existing document-level neural machine translation (NMT) models have sufficiently explored different context settings to provide guidance for target generation. However, little attention is paid to inaugurate more diverse contexts for abundant context information. In this paper, we propose a Selective Memory-augmented Neural Document Translation model (SMDT) to expand the context and reduce the hypothesis space of attention. Specifically, we retrieve similar bilingual sentence pairs from the training corpus to augment global context and then extend the two-stream attention model with a selection mechanism to capture local context and diverse global contexts. This unified approach allows our model to be trained elegantly on three publicly document-level machine translation datasets and significantly outperforms previous document-level NMT models.