Sentiment Analysis of Twitter using Machine Learning

Abstract

Social media today makes a shift in lifestyle of many people. Twitter is often used for giving campaigns, critics and opinions that can make pros and cons. So, there are large amounts of textual data contained in twitter called big data. We can crawl the Twitter data and use it for Sentiment analysis to predict positive, negative or neutral sentiment. Finding the best combination algorithms is the key to success in sentiment analysis. Therefore, we compare the combination algorithms of preprocessing, feature extraction, feature selection and classification method. The research framework takes an unique approach to tweet information by combining an approved polarities lexicon learned from customer reviews of domains with tweet-specific characteristics and unigrams to create a classification model employing machine learning methods.