Can QE-informed (Re)Translation lead to Error Correction?

TL;DR

This paper explores QE-informed retranslation methods for error correction in machine translation, demonstrating that a simple, training-free approach can outperform traditional APE systems in quality evaluation tasks.

Contribution

It introduces a training-free QE-informed retranslation method that effectively selects or edits translations, outperforming existing approaches in error correction for MT.

Findings

The first approach achieved a Delta COMET score of 0.0201.

The second approach achieved a Delta COMET score of -0.0108.

The first approach won the subtask leaderboard.

Abstract

The paper presents two approaches submitted to the WMT 2025 Automated Translation Quality Evaluation Systems Task 3 - Quality Estimation (QE)-informed Segment-level Error Correction. While jointly training QE systems with Automatic Post-Editing (APE) has shown improved performance for both tasks, APE systems are still known to overcorrect the output of Machine Translation (MT), leading to a degradation in performance. We investigate a simple training-free approach - QE-informed Retranslation, and compare it with another within the same training-free paradigm. Our winning approach selects the highest-quality translation from multiple candidates generated by different LLMs. The second approach, more akin to APE, instructs an LLM to replace error substrings as specified in the provided QE explanation(s). A conditional heuristic was employed to minimise the number of edits, with the aim of maximising the Gain-to-Edit ratio. The two proposed approaches achieved a Delta COMET score of 0.0201 and -0.0108, respectively, leading the first approach to achieve the winning position on the subtask leaderboard.