Emotion Classification of Duterte Administration Tweets Using Hybrid Approach
Randell Gimenez, Melvin Gaviola, Mary Jane G. Sabellano, Ken Gorro · 2017
Nowadays, people engage in social networking sites to gain information, share thoughts and ideas, and give reaction to certain topics. The works of Duterte Administration has been a hot topic in the country since the day it started its term. This research has gathered tweets and determined the emotions based on the Ekman's six basic emotion classification namely happiness, sadness, anger, fear, disgust, and surprise. A hybrid approach was used to determine the classification using lexicon-based and machine learning. Support Vector Machine (SVM) and Naïve Bayes Classifier (NBC) were used to train the classifier. Result shows that SVM gains a higher accuracy of 80.48% over NBC. Most tweets result to anger and a few express surprises as classification of emotions.