Enhancing Arabic Sentiment Analysis Through Machine Learning and Deep Learning Techniques

This study investigates the effectiveness of various machine learning techniques for sentiment analysis, focusing on data from multiple social media platforms. A comparative analysis was performed between traditional machine learning algorithms such as Naïve Bayes (NB), Support Vector Machines (SVM), and Random Forest (RF), alongside the Long Short-Term Memory (LSTM) recurrent neural network. Our results demonstrate that the LSTM approach significantly outperforms the traditional methods, achieving a classification accuracy of <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathbf{9 2. 1 3 \%}$</tex>. The research provides foundational insights for future studies in Arabic sentiment analysis, particularly in understanding consumer behavior across different sectors. The adopted methodology for extracting and analyzing consumer opinions emphasizes the critical role of data preprocessing in improving dataset quality and system accuracy. Limitations identified include the misclassification of neutral sentiments and the need for a multi-class classification framework. Future work aims to broaden the dataset to over 25,000 posts and explore multilingual sentiment analysis, addressing the complexities of classifying Arabic and English posts while expanding the model's application to additional natural language processing tasks.