Year: 2025
Type: article
Abstract: Sentiment analysis has become an integral part of natural language processing, especially in social media, where large volumes of user-generated content are readily available. This review delves into the various deep learning techniques used in sentiment analysis, assessing their advantages and drawbacks across multiple languages and settings. It examines document-level, sentence-level, and aspect-based sentiment analysis methodologies, emphasizing the progress made with models like Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and particularly BERT and its adaptations. Moreover, the paper discusses the existing challenges in sentiment classification, including the complexities of sarcasm detection, multilingual processing issues, and the importance of effective preprocessing techniques. The findings highlight the significance of sentiment analysis in diverse fields, including education, brand management, finance, and emergency response. Ultimately, this review identifies opportunities for future research, such as the integration of advanced models, the inclusion of underrepresented languages, and the development of interpretable frameworks to enhance trust in sentiment analysis applications.
Source: Journal of Neonatal Surgery
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: diamond
OpenAlex ID: W7219509014