Sentiment Analysis Using Machine Learning Classifiers: Evaluation of Performance

The modern era internet is full of information pertaining to opinion of people through applications such as social media, micro blogging sites, review sites, personal blogs etc. Sentimental analysis is an area in text mining where opinion of people can be analyzed and classified into positive, negative or neutral. In this work, the sentiments of the tweets or reviews published in the twitter is identified by searching for the particular keyword in tweets and then evaluate the polarity of the tweets as positive and negative. The sentiments of the tweets that are tweeted on a twitter evaluated based on feature selection of each score words. In order to select the best features Naive Bayes Classifier (NBC) is used for training and testing the features of a words and also evaluating the sentiment polarity of each tweets. Performance evaluation parameters such as accuracy, precision and time is taken into consideration and compared with three machine learning classifiers, namely, Random Forest, Naive Bayes and Support Vector Machine(SVM).