Improving Lexically Constrained Neural Machine Translation with Source-Conditioned Masked Span Prediction

TL;DR

This paper introduces a novel training strategy for lexically constrained neural machine translation that significantly improves translation accuracy of domain-specific, long, and specialized terms by leveraging source-conditioned masked span prediction.

Contribution

The paper presents a new training approach based on masked span prediction that enhances the handling of long, domain-specific terms in lexically constrained NMT.

Findings

Consistent improvements on terminology accuracy

Enhanced sentence-level translation quality

Effective across multiple domain-specific corpora and language pairs

Abstract

Accurate terminology translation is crucial for ensuring the practicality and reliability of neural machine translation (NMT) systems. To address this, lexically constrained NMT explores various methods to ensure pre-specified words and phrases appear in the translation output. However, in many cases, those methods are studied on general domain corpora, where the terms are mostly uni- and bi-grams (>98%). In this paper, we instead tackle a more challenging setup consisting of domain-specific corpora with much longer n-gram and highly specialized terms. Inspired by the recent success of masked span prediction models, we propose a simple and effective training strategy that achieves consistent improvements on both terminology and sentence-level translation for three domain-specific corpora in two language pairs.