Enhanced Social Media Opinion Mining Using Feature Optimization and Ensemble Machine Learning

Abstract

With the increasing amount of user generated content (UGC) on social media, accurate and efficient opinion mining has become more and more crucial. In this research, an improved framework for sentiment classification with feature optimization and ensemble machine learning technique is explored. Natural language processing (NLP) methods are used to process textual data and represent data as Term Frequency – Inverse Document Frequency (TF-IDF). Redundant attributes were thought to be eliminated by feature-selection methods such as Information Gain, Chi-square or optimization using the genetic algorithm. Individual performance of Support Vector Machine, Random Forest, Logistic Regression and Naive Bavies are examined as well as their performance in ensemble. Evaluation models include accuracy, precision, recall, F1-score and hypothesis testing. The suggested framework strives to enhance the reliability of sentiment classification, construction of lower dimensionality feature space, and the robustness of social media sentiment classification for practical use.