Diagnose, Then Repair: A Two-Stage MQM-Guided Post-Editing Framework for Domain-Specific Machine Translation

2026-09-24T16:13:45.287Z

• summary: Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 6: Industry Track) → LLM-based machine translation evaluation can closely match human judgments, but in practice it remains largely diagnostic, with the signals rarely translating into direct quality improvements under real production constraints. We propose a two-stage, evaluator-guided automatic post-editing framework that turns MQM-style evaluation into targeted repairs: a retrieval-augmented LLM evaluator outputs structured, span-level MQM diagnoses under an explicit edit contract, and a separate LLM post-editor applies minimal edits restricted to those diagnoses. This separation improves controllability and r
• publishedAt: Not provided → 2026-09-19T05:52:07.000Z
• url: https://aclanthology.org/2026.acl-industry.115/ → https://arxiv.org/abs/2609.22793