An Exploration of Neural Sequence-to-Sequence Architectures for Automatic Post-Editing

Abstract

In this work, we explore multiple neural architectures adapted for the task of automatic post-editing of machine translation output. We focus on neural end-to-end models that combine both inputs $mt$ (raw MT output) and $src$ (source language input) in a single neural architecture, modeling $\{mt,src\}\to pe$ directly. Apart from that, we investigate the influence of hard-attention models which seem to be well-suited for monolingual tasks, as well as combinations of both ideas. We report results on data sets provided during the WMT-2016 shared task on automatic post-editing and can demonstrate that dual-attention models that incorporate all available data in the APE scenario in a single model improve on the best shared task system and on all other published results after the shared task. Dual-attention models that are combined with hard attention remain competitive despite applying fewer changes to the input.

Accepted for presentation at IJCNLP 2017

References in corpus (2)

• Is Neural Machine Translation Ready for Deployment? A Case Study on 30 Translation Directions
• Morphological Inflection Generation with Hard Monotonic Attention

Cited by in corpus (6)

• eSCAPE: a Large-scale Synthetic Corpus for Automatic Post-Editing
• Asynchronous Bidirectional Decoding for Neural Machine Translation
• Improving Multi-turn Dialogue Modelling with Utterance ReWriter
• Retrieve and Refine: Improved Sequence Generation Models For Dialogue
• LIG-CRIStAL System for the WMT17 Automatic Post-Editing Task
• Think Again Networks and the Delta Loss