!MSA at BAREC Shared Task 2025: Ensembling Arabic Transformers for Readability Assessment

Résumé

We present !MSA's winning system for the BAREC 2025 Shared Task on fine-grained Arabic readability assessment, achieving first place in six of six tracks.Our approach is a confidence-weighted ensemble of four complementary transformer models (AraBERTv2, AraELECTRA, MARBERT, and CAMeL-BERT) each fine-tuned with distinct loss functions to capture diverse readability signals.To tackle severe class imbalance and data scarcity, we applied weighted training, advanced preprocessing, SAMER corpus relabeling with our strongest model, and synthetic data generation via Gemini 2.5 Flash, adding 10k rarelevel samples.A targeted post-processing step corrected the prediction distribution skew, delivering a 6.3% Quadratic Weighted Kappa (QWK) gain.Our system reached 87.5% QWK at the sentence level and 87.4% at the document level, demonstrating the power of model and loss diversity, confidence-informed fusion, and intelligent augmentation for robust Arabic readability prediction. 1

Citer ce document

Accès au document

Texte intégral en lecture en ligne, réservé aux abonnés SPHAERO et aux membres de l'institution. Se connecter

Auteur(s)

Statistiques

Consultations : 1

Téléchargements : 0