Optimizing Sentiment Analysis in Social Media with a Machine Learning Approach using Naive Bayes and Support Vector Machines

This paper offers detailed comparative analysis of the most popular approaches to machine learning, i.e., Naive Bayes (NB) and Support Vector Machines (SVM), employed in sentiment analysis on social networks. This study is based on an eclectic dataset which contains the information from different popular social media sites that includes topics, sentiments and language styles etc. The simple and popular Naive Bayes algorithm is contrasted to the Support Vector Machines, which are more sophisticated but effective. These two algorithms are realized and fine-tuned to enhance their effectiveness in capturing subtle sentiments from short and informal content that is posted on social media. The findings indicate that although Naive Bayes achieves excellent performance and competitive efficiency, Support Vector Machines present better accuracy in revealing the slight differences of subtleties and contextual sentiments. The findings contribute to the body of literature on sentiment analysis approaches and indicate how Naive Bayes and Support Vector Machines work on social data. The findings of such a research are also linked to the development of more reliable sentiment analysis tools that target social media applications. This research further contributes to a better understanding of the feature discovery capabilities of these ML techniques in relation to sentiment analysis tasks, helping practitioners and researchers select appropriate models for changing online conversation settings.