Feature Selection and Polarity Classification Using Machine Learning Algorithms NB & SVM

Abstract

Sentiment analysis and its classification of social data has become challenging now a days because of unstructured nature of data, slang, misspells and abbreviations used by customers while giving comments or reviews. Using machine learning approach for sentiment analysis helps in finding useful patterns and derive predictions which are important in decision making for improvement of overall products and customer satisfaction. In this paper we use tweets for famous mobile brands like Iphone, Vivo and Red MI. Machine learning algorithm like naive Bayes and SVM are used to find polarity of tweets like positive, negative or neutral. This helps to find popular brands. Also we compare overall accuracy of these algorithms using measures like precision and recall and f measure.