Sentiment Analysis Optimization Using Hybrid Machine Learning Techniques
Ravindra Changala, H. N. Lakshmi, Venkata Krishna Gandikota, H Bhagya Lakshmi, G Uday Kiran, Venkata Nagaraju Thatha · 2024
Researchers are driven to complete their work in sentiment analysis by the ever-increasing demands placed on government agencies and commercial companies. The way people express themselves on social media is a reflection of their opinions on various products, services, and events. As a subfield of NLP, sentiment analysis aims to extract positive or negative polarities from text found in social media platforms. Three cutting-edge machine learning classifiers-Naive Bayes, SVM, and OneR-are showcased in this study for the purpose of optimizing sentiment analysis. Two benchmark datasets are used in the studies; one dataset is derived from Amazon, while the other is derived from IMDB movie reviews. We compare and analyze the results of these classification methods. The Naive Bayes learned rather quickly, but OneR shows more promise with a precision of 92.6%, an F-measure of 96%, and a properly categorized occurrence rate of 93.4%.