Reproduce “!MSA at BAREC Shared Task 2025: Ensembling Arabic Transformers for Readability Assessment” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Mohamed Basem, Mohamed Younes, Seif Ahmed, Abdelrahman Moustafa

The authors build !MSA to assign Arabic sentences and documents to 19 ordered readability levels. It combines confidence-weighted predictions from Arabic transformers using classification, regression, and ordinal regression, with track-specific data handling. On the official sentence-level strict BAREC track, it achieves 87.5% QWK versus an 84.6% baseline; this is benchmark evidence, not human validation or evidence of broader generalization.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.