In plain words: A system that fixes machine-translated German by reading the surrounding sentence for context, trained alongside a task that teaches it to clean up corrupted text. It was entered in the 2019 English-German post-editing competition; the paper describes the setup without reporting how it compared.

Abstract

In this paper, we describe our submission to the English-German APE shared task at WMT 2019. We utilize and adapt an NMT architecture originally developed for exploiting context information to APE, implement this in our own transformer model and explore joint training of the APE task with a de-noising encoder.

Hongfei Xu, Qiuhui Liu, Josef van Genabith
arXiv:1908.03402 · cs.CL · submitted Aug 9, 2019
abstract · pdf · html · WMT 2019 Automatic Post-Editing Shared Task Paper