SENTIMENT ANALYSIS OF SOCIAL MEDIA DATA UTILIZING NATURAL LANGUAGE PROCESSING AND MACHINE LEARNING TECHNIQUES

Abstract

The swift expansion of social media has generated a vast quantity of unstructured textual data that mirrors public sentiment on diverse subjects. Examining this data yields significant insights for enterprises, governments, and scholars. This research seeks to create a sentiment analysis system utilizing Natural Language Processing (NLP) and machine learning techniques to categorize social media messages as positive, negative, or neutral sentiments. The proposed system comprises several essential stages: text preprocessing, feature extraction via Term Frequency–Inverse Document Frequency (TF-IDF), and classification employing machine learning methods including Naïve Bayes, Support Vector Machine (SVM), and Logistic Regression. A dataset including 10,000 social media postings was meticulously collected and extensively annotated to guarantee precision in sentiment classification. Experimental results indicated that SVM attained superior performance, achieving an accuracy of 87.4% and an F1-score of 0.86, surpassing both Naïve Bayes and Logistic Regression. The results illustrate the efficacy of natural language processing integrated with machine learning in the analysis of extensive social media datasets, providing a reliable method for sentiment classification. The study underscores the efficacy of sentiment analysis in gauging public opinion, facilitating commercial decisions, and identifying nascent social trends.