TTLab at AlexandriaX-2026: A Fine-Tuned Surface Tagger for Arabic Machine-Translation Error-Span Detection and Classification

2026-09-24

AI Summary

This study addresses the challenges of label imbalance and rare category recognition in Arabic machine translation error detection and classification. The task is formulated as token-level classification with preserved character offsets to optimize error localization. Methodologically, a MARBERTv2 pre-trained encoder is fine-tuned, incorporating a class-weighted focal loss and dialect-adaptive decoding thresholds to effectively mitigate the long-tail data distribution problem. The proposed approach achieves scores of 40.8 and 40.91 on the development and test sets, respectively, ranking third in the competition. These results demonstrate significant improvements in the localization and classification of cross-dialectal error spans.

Abstract

We present TTLab's submission to the AlexandriaX-2026 Subtask~3 on Arabic MT error span detection and classification. Our system frames the task as token-level classification over surface forms, preserving character offsets to ensure exact alignment with the evaluation metric. To handle severe label imbalance, we employ a focal loss with class weighting and dialect-specific decoding thresholds. Among six Arabic pre-trained encoders, MARBERTv2 achieves the best overall performance of 40.8 and 40.91 on the development and test set, respectively, ranking $\nth{3}$ out of all participating teams. While our system localizes error spans effectively, classification of rare error types remains challenging, highlighting the need for data augmentation for tail categories. The code is available at ${\href{https://github.com/ENTAILab/arabic-dialectal-mt-error-span-detection}{\faGithub~ TTLab at AlexandriaX-2026}$

Problem

• Arabic machine translation
• error span detection
• error classification
• label imbalance

Innovation

• Token-level Classification
• Focal Loss
• Surface Tagger
• Error Span Detection
• MARBERTv2