An analysis of different sequential pattern based sentimental analysis techniques using product reviews

Sentiment analysis is the process of finding opinions or feelings about a specific topic in written text. With the rise of social media, more people are sharing their thoughts online. This has made sentiment analysis very important because the opinions found in the text can help with things like recommendation systems, understanding social networks, predicting market trends, and political discussions. A key part of this analysis is aspectterm extraction, especially from customer reviews. This means recognizing specific details or features mentioned in reviews, which can be quite challenging. This survey examines 25 research papers focused on sequential pattern-based sentiment analysis. It begins by outlining a common framework for sequential pattern-based sentiment analysis and categorizes it into five primary approaches: Attention mechanism-based, Transformer-based, Statistical, Machine Learning (ML)- based, Deep Learning (DL)-based, and sequential pattern- based approaches. The analysis of the existing techniques highlights research gaps that can guide future innovations in the field. This study reviews the methodologies employed in the 25 articles and also analyzes publication year, research approaches and performance metrics of various methodologies. The evaluations how that DL-based approaches are the most frequently employed methods. Moreover, Python emerges as a prevalent tool, while the most commonly utilized database is Amazon review database. Additionally, accuracy is the majorly used analytical metric, often reported with high values. Keywords: Product review, Sentiment analysis, sequential pattern-based Sentiment analysis, Machine learning Deep Learning.