Enhancing Sentiment Analysis through Supervised Machine Learning Techniques
Khushboo Jain, Arun Agarwal, Kamlesh Kumar Raghuvanshi, Sheetal Singh · 2025
Words and phrases reflect people’s perspectives on products, services, governments, and social events across various social media platforms. The discernment of positive or negative sentiments within this digital discourse is an indispensable facet of natural language processing, termed sentiment analysis. The rising demands from business entities and governmental bodies have spurred researchers to intensify efforts in enhancing sentiment analysis methodologies, aiming to extract nuanced insights that resonate with the evolving landscape of digital communication and the dynamic interplay of societal perspectives. This work utilizes cutting-edge machine learning (ML) classification algorithms like decision tree, Naïve Bayes, and one rule (One R) to optimize sentiment analysis. The experiments involve two manually compiled datasets, one from Amazon and another from Internet Movie Database (IMDB) movie reviews. The effectiveness of these supervised ML algorithms is examined and compared. Naïve Bayes exhibited a fast learning rate, while one R demonstrated promising results with an accuracy of 94.5% in precision, 93.2% in F -score, and 92.43% as correctly classified instances.