Predictive Modeling and Sentiment Classification of Social Media Through Extreme Learning Machine

摘要

Social media platform has revolutionized all sections of the society. The popularity of social or community netwroking platform in the last decade has created new opening to analyze and study public opinions and sentiments for use in financial and social behavioral studies. On the other hand, machine-learning techniques have also laid a significant impact. Machine learning approaches are widely implemented in processing and analyzing sentiments. Extreme Learning Machine is the most favoured machine leaning classifier, which shows better results apart from support vector machine classifier. The working principle of extreme learning machine can represent results in categorical form. However, one-to-one sentiment classification may not disclose too much information, which could have been beneficial for research purpose. So multi-class sentiment has been discussed here with the help of extreme learning machine. The experimental results show that extreme leaning machine achieves better accuracy and performance in comparison to other machine learning classifiers.