Twitter Sentiment Classification on Sanders Data using Hybrid Approach

Sentiment analysis is very perplexing and massive issue in the field of social data mining. Twitter is one of the mostly used social media where people discuss on various issues in a dense way. The tweets about a particular topic give peoples' views, opinions, orientations, inclinations about that topic. In this work, we have used pre-labeled (with positive, negative and neutral opinion) tweets on particular topics for sentiment classification. Opinion score of each tweet is calculated using feature vectors. These opinion score is used to classify the tweets into positive, negative and neutral classes. Then using various machine learning classifiers the accuracy of predicted classification with respect to actual classification is being calculated and compared using supervised learning model. Along with building a sentiment classification model, analysis of tweets is being carried out by visualizing the wordcloud of tweets using R.