Real time sentiment analysis of tweets using Naive Bayes

Abstract

Twitter1 is a micro-blogging website which provides platform for people to share and express their views about topics, happenings, products and other services. Tweets can be classified into different classes based on their relevance with the topic searched. Various Machine Learning algorithms are currently employed in classification of tweets into positive and negative classes based on their sentiments, such as Baseline, Naive Bayes Classifier, Support Vector Machine etc. This paper contains implementation of Naive Bayes using sentiment140 training data using Twitter database and propose a method to improve classification. Use of SentiWordNet along with Naive Bayes can improve accuracy of classification of tweets, by providing positivity, negativity and objectivity score of words present in tweets. For actual implementation of this system python with NLTK and python-Twitter APIs are used.