Enhancing Named Entity Recognition for Libyan Arabic Dialect Using Neural Network Models

Named Entity Recognition is a key technique in Natural Language Processing task that extracts entities such as persons, organizations or locations. It remains a significant challenge for under-resourced Arabic dialects, particularly Libyan Arabic dialect, due to its phonological, lexical and syntactic divergence from Modern Standard Arabic and the scarcity of annotated corpora. This paper presents the first dedicated Named Entity Recognition for Libyan Arabic dialect using deep learning models. An annotated dataset was curated from social media networks, and model performance was evaluated using precision, recall, and F1-score. The result achieved an overall accuracy of 0.85, with F1-scores of 0.88 (Location), 0.80 (Organization), and 0.79 (Person). Keywords: Named Entity Recognition, Libyan Arabic dialect, Deep Learning, Low-Resource Languages.