Work

ZAI at BAREC Shared Task 2025: AraBERT CORAL for Fine Grained Arabic Readability

Year: 2025

Type: conference-paper

Abstract: Readability assessment is essential for effective communication of scientific and medical content in Arabic.We present a system for the BAREC 2025 Shared Task Arabic Readability Assessment.The system fine-tunes AraBERTv2 with a CORAL ordinal head, applies AraBERT-specific preprocessing, and selects checkpoints using Quadratic-Weighted Kappa (QWK) with early stopping.Our model achieves a QWK of 85.5 on the Sentence Blind Test, demonstrating its effectiveness for automatic Arabic readability prediction.

Source: Arabic Natural Language Processing Workshop (WANLP)

DOI: 10.18653/v1/2025.arabicnlp-sharedtasks.37

Author: Ahmad M. Nazzal

Institution: Zayed University

Language: English

Topic: Text Readability and Simplification

Subfield: Artificial Intelligence

Field: Computer Science

Domain: Physical Sciences

Open Access status: hybrid

OpenAlex ID: W4416037814

FWCI:

Cites:

Cited by:

Related to: 10