Improved Data Augmentation for Translation Suggestion

TL;DR

This paper presents an improved translation suggestion system utilizing ensemble models, synthetic data construction, and multi-phase pre-training, achieving high rankings in the WMT'22 shared task.

Contribution

It introduces a novel combination of ensemble architectures, synthetic data strategies, and multi-phase pre-training for translation suggestion.

Findings

Ranked second in English-German task

Ranked third in English-Chinese task

Effective use of synthetic data and multi-phase pre-training

Abstract

Translation suggestion (TS) models are used to automatically provide alternative suggestions for incorrect spans in sentences generated by machine translation. This paper introduces the system used in our submission to the WMT'22 Translation Suggestion shared task. Our system is based on the ensemble of different translation architectures, including Transformer, SA-Transformer, and DynamicConv. We use three strategies to construct synthetic data from parallel corpora to compensate for the lack of supervised data. In addition, we introduce a multi-phase pre-training strategy, adding an additional pre-training phase with in-domain data. We rank second and third on the English-German and English-Chinese bidirectional tasks, respectively.