Emotion Detection Framework for Twitter Data Using Supervised Classifiers

Abstract

“The task of emotion detection usually involves the analysis of text. Humans show universal consistency in identifying emotions however shows an excellent deal of variation between individuals in their abilities.” We have detected the emotion for Twitter messages as they provide rich ensemble of human emotions. We have used machine learning algorithms namely Naive Bayes (NB) and k-nearest neighbor algorithm (KNN) to detect the emotion of Twitter message and then classify the Twitter messages into four emotional categories. We also made a comparative study of two supervised machine learning algorithms; the eager learning classifier (NB) performed well when compared with lazy learning classifier (KNN).