Papers by Mohanad Mohamed
Enhancing Arabic NLP Tasks through Character-Level Models and Data Augmentation (2025.coling-main)
Copied to clipboard
| Challenge: | Using character-level models, natural language processing for Arabic is challenging due to its rich morphology, root-based word formation, flexible sentence structures, diacritical ambiguities, and orthographic variations. |
| Approach: | They propose a character-level approach specifically designed for Arabic NLP tasks that incorporates Convolutional Neural Networks (CNNs), pre-trained transformers (CANINE), and Bidirectional Long Short-Term Memory networks (BiLSTMs). |
| Outcome: | The proposed model outperforms existing models on Arabic privacy policy classification task and reports a micro-averaged F1 score of 93.8%, surpassing state-of-the-art models. |