Classification of Indonesian Presidential Campaign on Twitter Using Word2Vec

Abstract

Indonesia is a country that use democracy system and one of the largest in the world. The process of democracy continues every 5 years and always raises the pros and cons in the community especially on social media. This research explains the process of presidential election in Indonesia using Word2Vec as an extraction feature. Some classifications used are K-Nearest Neighbors (K-NN), Naïve Bayes, Decision Tree, Random Forest, and Support Vector Machine (SVM) to decide sentiment. The purpose of this research is to compare several classifications as well as to know the highest accuracy of some classification use. Post data about presidential candidates through the crawling process on social media Twitter. Data originating directly from the community and national media produce variations of various responses. The data processed is as much as 640 in Bahasa Indonesia with the keywords Prabowo, Sandi, Jokowi, and Ma'ruf. The results showed the highest accuracy gained when using the Random Forest Classification method, with the highest accuracy reaching 98.33% and lowest accuracy reaching 81.96% using Support Vector Machine.