Navigating Sentiment Analysis Horizons: A Comprehensive Survey on Machine Learning Approaches for Unstructured Data in Medical Sciences and Science and Technology

Abstract

The increasing field of sentiment analysis on unstructured data has become a focal point of research, witnessing a proliferation of machine learning techniques. This extensive survey investigates into the methodologies embraced by researchers across diverse domains, spotlighting the pivotal role played by automatic feature learning and word embedding models in the booming success of sentiment analysis approaches. The exploration of sentiment analysis techniques, the survey unravels the workings of Support Vector Machine (SVM), Naive Bayes (NB), Artificial Neural Networks (ANN), Decision Trees (DT), K Nearest Neighbour (k-NN), Random Forest (RF), and metaheuristic optimization algorithms, elucidating their time complexities, advantages, and limitations. By synthesizing the challenges faced by researchers, the survey not only offers prominent insights but also depicts the course for future investigations, presenting an open issue in the sentiment analysis. The discourse extends beyond theoretical considerations to practical
applications, evaluating the performance of sentiment analysis techniques across a spectrum of real-world datasets. As a comprehensive resource, this survey provides researchers and practitioners with a understanding of the evolving paradigm, fostering informed decision-making and inspiring future innovations in sentiment analysis on unstructured data within the paradigm of machine learning.