ALSEREIDI, MOHAMED SOHAIL2025-01-242025-01-242023-0721003440https://bspace.buid.ac.ae/handle/1234/2762This study aimed to explore the effectiveness of deep learning models, including Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Transformer-based models, and Graph Neural Networks (GNNs), in aspect-based sentiment analysis (ABSA) of textual opinions. The research conducted a comprehensive literature review to analyse existing studies and articles in the field, focusing on comparing the performance of deep learning models with traditional rule-based or lexicon-based approaches. The findings indicated that deep learning models demonstrated promising results and surpassed the performance of traditional methods in ABSA. CNN-based models, in particular, achieved remarkable outcomes on benchmark datasets. Transformer-based models, such as BERT and RoBERTa, also exhibited strong performance across various natural language processing tasks, including sentiment analysis. Additionally, GNNs showcased potential in leveraging text structure to improve aspect and sentiment extraction. The research identified a research gap, emphasizing the need for further exploration and advancements in the utilization of deep learning models for ABSA. This study contributes to the understanding of the potential of deep learning in ABSA and provides insights for future research in this field.enaspect-based sentiment analysis, deep learning, sentiment analysis, natural language processing, Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), Graph Neural Networks (GNNs), rule-based methods, textual opinionsDissertation