Wathiq-SPU at AraSeg Shared Task 2026: Calibrated Transformer Ensembles for Arabic Sentence Segmentation across Punctuation and Paragraph Conditions
Summary
Arabic sentence segmentation becomes substantially more difficult when punctuation and paragraph boundaries are unavailable. The AraSeg Shared Task 2026 provides a controlled benchmark for studying this problem under four subtasks that progressively remove str...
Arabic sentence segmentation becomes substantially more difficult when punctuation and paragraph boundaries are unavailable. The AraSeg Shared Task 2026 provides a controlled benchmark for studying this problem under four subtasks that progressively remove structural information. We present the Wathiq-SPU submission to all four subtasks of the Closed Track: Paragraph-Aware (PA), No-Punctuation Paragraph-Aware (NoPnx-PA), No-Paragraph (NP), and No-Punctuation No-Paragraph (NoPnx-NP).We formulate sentence segmentation as token-level binary boundary classification using pretrained Arabic Transformer encoders and combine calibrated predictions with subtask-specific ensemble strategies that adapt to the amount of available structural information. CAMeLBERT-MSA serves as our primary encoder, while the most challenging NoPnx-NP subtask additionally benefits from AraELECTRA-based backbone diversity. Our experiments reveal that the most effective source of model diversity depends on task difficulty: diversity among independently trained models provides robust gains when some structural information is available, whereas backbone diversity becomes increasingly important when both punctuation and paragraph cues are removed. The selected systems achieve F1 scores of 93.4, 87.9, 90.0, and 84.8 on the PA, NoPnx-PA, NP, and NoPnx-NP subtasks, respectively.
Researchers
- Riad Sonbol Co-author