A Comprehensive Survey on Sentiment Analysis and Opinion Mining on Social Media Using Machine Learning Techniques
Main Article Content
Abstract
The growing usage of Internet-based applications, particularly social media platforms and blogs, resulted in an unprecedented flow of thoughts, reviews, and opinions. Sentiment Analysis (SA), also known as opinion mining, has emerged as a vital technique for systematically gathering and assessing people's sentiments, opinions, and impressions on a wide range of subjects, whether they are products, themes, or services, in this age of digital connectivity. The variety of public mood data has shown to be a valuable resource for corporations, governments, and individuals, enabling sound decision-making. However, impediments to precise judgment of sentiment polarity and precise interpretation of feelings develop as a result of SA implementation. SA is a sophisticated science that identifies and extracts subjective information from textual material. This is performed by utilizing strong Natural Language Processing (NLP) and text mining techniques. Our post aims to provide a thorough overview of the complicated procedures that underpin SA, as well as an examination of its problems. The SA process begins with data gathering from publically available datasets, followed by a number of data pre-processing activities such as converting text to lowercase, dealing with contractions, tokenizing, removing short and repeated words, and so on. Following that, feature extraction, including content-based, document-based, and texture-based features, is examined. Following that, the technique investigates Feature Selection (FS) approaches such as filter, wrapper, embedding, and hybrid procedures. Finally, we investigate numerous classification methods for sentiment detection, including machine learning, lexicons, and hybrid approaches. By presenting an in-depth review of key SA techniques and procedures, this study presents a comprehensive evaluation report that can serve as a solid foundation for future studies.