Enhancing Arabic Text Readability Assessment: A Combined BERT and BiLSTM Approach

Abstract

Modern readability assessment methods employ innovative approaches such as natural language processing and deep learning techniques. These techniques facilitate a comprehensive analysis of text, focusing on elements like context, syntax, and semantics. Such advanced methods are highly effective in adapting texts to suit learners’ abilities. They also enhance text evaluation by systematically classifying textual content into different readability levels. This paper introduces a novel approach that uses a hybrid BERT-BiLSTM model for Arabic readability assessment, representing a significant advancement in the classification of text difficulty. The highest level of accuracy achieved with this model is $89.53 \%$.