SMDT: Selective Memory-Augmented Neural Document Translation

TL;DR

This paper introduces SMDT, a neural document translation model that enhances context diversity by selectively retrieving and integrating global and local contexts, leading to improved translation quality.

Contribution

It proposes a novel selective memory-augmented approach that retrieves similar sentence pairs to enrich context, outperforming previous document-level NMT models.

Findings

Significant performance improvements over previous models.

Effective integration of global and local context.

Successful training on multiple datasets.

Abstract

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 context for abundant context information. In this paper, we propose a Selective Memory-augmented Neural Document Translation model to deal with documents containing large hypothesis space of the context. Specifically, we retrieve similar bilingual sentence pairs from the training corpus to augment global context and then extend the two-stream attention model with selective 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.