Twitter sentiment analysis with different feature extractors and dimensionality reduction using supervised learning algorithms

Twitter is an online micro-blogging platform which allows us to treasure trove about the current circumstance at any juncture in time. In this paper, we analyze the sentiments of huge amount of tweets generated from Twitter users which are stored in Twitter database. We have chosen accuracy as the evaluation criteria of classification methods, namely Naive Bayes and Support Vector Machine. Also we have used unigram and bigram as feature extractors along with Chi2 and Singular Value Decomposition for dimensionality reduction. Through tokenization, having several stages of pre-processing and several combinations of feature vectors and classification methods, we are able to achieve an accuracy of 89.61% when analyzing the sentiment of tweets.