Structured Document Translation via Format Reinforcement Learning

TL;DR

This paper introduces FormatRL, a reinforcement learning approach that enhances document translation by directly optimizing structure-aware rewards, leading to improved structural and translation quality in complex document formats.

Contribution

The paper presents FormatRL, a novel reinforcement learning framework that incorporates structure-aware rewards for better document-level translation of XML/HTML structures.

Findings

Improved performance across six evaluation metrics.

Effective optimization of structural similarity and translation quality.

Analysis of reward functions' impact on translation and structure.

Abstract

Recent works on structured text translation remain limited to the sentence level, as they struggle to effectively handle the complex document-level XML or HTML structures. To address this, we propose \textbf{Format Reinforcement Learning (FormatRL)}, which employs Group Relative Policy Optimization on top of a supervised fine-tuning model to directly optimize novel structure-aware rewards: 1) TreeSim, which measures structural similarity between predicted and reference XML trees and 2) Node-chrF, which measures translation quality at the level of XML nodes. Additionally, we apply StrucAUC, a fine-grained metric distinguishing between minor errors and major structural failures. Experiments on the SAP software-documentation benchmark demonstrate improvements across six metrics and an analysis further shows how different reward functions contribute to improvements in both structural and translation quality.