Sentiment Analysis and Text Classification for Social Media Contents Using Machine Learning Techniques

Abstract

utomated text classification has been considered a vital approach to manage and process unstructured documents generated from social media and plain text in digital forms. Machine learning techniques are found to be worthy of text mining. This article describes the various process such as pre-processes, analyzes, and visualizes social media contents to perform sentiment analysis using popular machine learning algorithms (Naive Bayes, Support Vector Machines, Decision Trees, Random Forest, AdaBoost, and Gradient Boosting) for analyzing online reviews. The classifiers are trained on a benchmark dataset and performance is assessed in terms of precision, recall, and accuracy. The significance of the individual method is empirically illustrated quantitatively. A set of systematic experiments are conducted on social media contents extracted from Kaggle. Experimental results indicate that the Support Vector Machine (SVM) method outperforms in terms of accuracy compared with Naive Bayes. Applications of sentiment analysis would make it easier for the business competitors to understand the views of stakeholders leading to increase revenues and help them in policy-making decisions.

Keywords: Text Mining, Social Media, Sentiment Analysis, Feature Selection, Machine Learning