State-of-the-Art Arabic Text Classification: Hybrid Feature Engineering and Ensemble Learning for Superior NLP Performance

Text classification is a fundamental task in Natural Language Processing (NLP) with applications in sentiment analysis, spam filtering, and document categorization. Despite significant advancements, Arabic text classification presents unique challenges due to the language’s complex morphology, diacritics, and dialectal variations. This paper explores the effectiveness of multiple Machine Learning (ML) models for Arabic text classification, including Logistic Regression (LR), Support Vector Machine (SVM), Naïve Bayes (NB), Random Forest (RF), Light Gradient Boosting Machine (LGBMClassifier), and Stochastic Gradient Descent Classifier (SGDClassifier). Additionally, two Ensemble Learning (EL) models (SVM + LGBMClassifier + LR) and (LGBMClassifier + LR) are employed to enhance classification performance. Our experimental results demonstrate that ensemble techniques outperform individual classifiers, achieving superior accuracy and robustness in Arabic text classification. This study provides valuable insights into the effectiveness of different ML models and feature extraction techniques for Arabic NLP, contributing to the development of more efficient Arabic text classification systems.