Clustering and Classification Based Approach for Emotion Analysis in Online Social Network Data

Social media plays a significant role in explore opinions and emotions of the users based on the day-to-day activities. The data mining techniques for social media data analysis and emotion mining based analysis are ranging from unsupervised to semi supervised and supervised learning methods. Mining of social network data about users opinion and emotion is necessary to understand the users behaviour and mentality. This research work proposed hybrid data mining approach using K-Means clustering and Naive Bayes classification techniques to analyse the emotions in the tweets. In the reprocessing phase, K-Means clustering process performed in the tweet emotion data set using Euclidean distance as distance function. The clustered data classified in the classification phase using Naïve Bayes classifier with 10 fold cross validation. Emotion type and cluster attributes used as the class variables to classify the clustered tweet emotion data. The experimental results shows Naïve Bayes classification in cluster data produces higher accuracy than the classification of tweet emotion data set without clustering.