Learning Non-Monotonic Automatic Post-Editing of Translations from Human Orderings

TL;DR

This paper investigates non-monotonic orderings in neural machine translation post-editing, analyzing human editing behaviors and training models to improve translation correction by leveraging human-like ordering patterns.

Contribution

It introduces a method to analyze human post-editing orderings and trains a system based on these insights, comparing it to traditional left-to-right and random orderings.

Findings

Humans tend to follow nearly left-to-right orderings.

Humans prefer starting with punctuation or verbs during editing.

Models trained on human orderings outperform others.

Abstract

Recent research in neural machine translation has explored flexible generation orders, as an alternative to left-to-right generation. However, training non-monotonic models brings a new complication: how to search for a good ordering when there is a combinatorial explosion of orderings arriving at the same final result? Also, how do these automatic orderings compare with the actual behaviour of human translators? Current models rely on manually built biases or are left to explore all possibilities on their own. In this paper, we analyze the orderings produced by human post-editors and use them to train an automatic post-editing system. We compare the resulting system with those trained with left-to-right and random post-editing orderings. We observe that humans tend to follow a nearly left-to-right order, but with interesting deviations, such as preferring to start by correcting punctuation or verbs.