This paper presents our submission to the Nakba-NLP 2026 Shared Task on binary image classification, where the goal is to categorize images of Gaza infrastructure as destroyed or intact. To address the challenges of class imbalance and resource-constrained deployment, we evaluated three convolutional architectures: ResNet50, MobileNetV2, and EfficientNet-B0, combined with a post-hoc threshold optimization step. Our results show that lightweight architectures are competitive with heavier models for this task, with EfficientNet-B0 achieving the highest Test F1-score of 0.85 despite having significantly fewer parameters than ResNet50. We further investigated the effect of input resolution, finding that increasing resolution improved ResNet50’s performance, though it remained below lightweight alternatives. Finally, we demonstrate that shifting the binary decision threshold from the default 0.50 to an optimized 0.45 improved ResNet50’s Test F1 from 0.79 to 0.81 by recovering recall for the minority destroyed class. Notably, this adjustment was only needed for ResNet50, while EfficientNet-B0 and MobileNetV2 performed best at the default 0.50, suggesting that larger models are more prone to majority-class bias. Overall, these results provide a systematic analysis of architectural efficiency and threshold behavior under class imbalance, offering practical insights for damage classification in resource-constrained crisis settings.

This paper presents our submission to the AbjadGenEval shared task on AI-generated text detection in Arabic and Urdu. To address the challenges of morphologically rich and low-resource environments, we developed a composite framework leveraging monolingual specialists (AraBERTv2, CAMeLBERT-DA) and multilingual transformers. Our system achieved robust in-domain performance with Test F1-scores of 0.75 for Arabic and 0.86 for Urdu. Methodologically, we tested both raw and normalized text to distinguish whether models detect based on semantic content or on surface artifacts such as punctuation and formatting patterns. Furthermore, our cross-lingual investigations reveal directional performance differences, where Urdu-trained models achieve 0.75 F1 on Arabic, while Arabic-trained models achieve only 0.61 F1 on Urdu. Despite this difference, both directions maintained notably high recall for the machine class, indicating that the model learns cross-lingual machine detection patterns across the Perso-Arabic script. Finally, transfer performance collapsed when internal layers were frozen, demonstrating that full fine-tuning is essential for cross-lingual detection. However, the observed performance differences may partly reflect data imbalance rather than purely linguistic factors.

This paper introduces resources for the computational study of scientific exegesis (Tafsir Ilmi): a structured ontology, a curated dataset of 194 scientifically relevant Quranic verses linked to 260 exegetical records from two authoritative Tafsir books, and an annotation framework that organizes scientific references by topic and sequential context. Existing Quranic resources treat verses as unstructured text, losing the logical order and causal relationships of scientific concepts documented in exegesis. To address this, we present QurSci-Onto, a three-layer ontology that categorizes verses by scientific domain, links them to authoritative Tafsir, and provides a framework for representing sequential processes through stage-based annotations. Our dataset includes page-level citations and covers 8 major scientific topics across 73 nodes. While the full corpus is tagged with broad categories and scientific topics, a specialized subset features granular node-level mappings to capture complex scientific narratives. We release QurSci-Onto as a foundational resource for Arabic semantic NLP and demonstrate that it enables significant improvements in semantic retrieval and enables multi-hop sequential reasoning capabilities over unstructured baselines.