Text

ANALISIS SENTIMEN PEMILIHAN PRA PEMILIHAN PRESIDEN TAHUN 2019 APLIKASI TWITTER DENGAN GEOLOCATION MENGGUNAKAN METODE NAÏVE BAYESIAN CLASSIFICATION

Share :
• 
• 
• 
• 
• 
• 
• 

ABSTRACT INDONESIAN PRESIDENTIAL PRESIDENTIAL ELECTION SELECTION ANALYSIS OF 2019 TWITTER APPLICATION WITH GEOLOCATION USING NAÏVE BAYESIAN CLASSIFICATION METHOD Wiranto Horsen Silitonga 1581017 2019 Indonesian Presidential Election is crowded to be discussed in the real world besides cyberspace, specifically on Twitter social media. Everyone is free to agree on the 2019 Indonesian Presidential candidate pair. Opinion raises many opinions, not only positive or neutral opinions but there are also negative opinions. Twitter's exclusive social media is now one of the most effective and efficient promotional or campaign venues to attract supporters. In this case, the researcher will conduct research on community leaders who are running for the presidency of Indonesia. The research method used in this study is the Naïve Bayesian Classifier classification algorithm. The data used are Indonesian tweets with Jokowi (# Jokowi2Periode) and Prabowo (#PrabowoSandi) keywords totaling 1018 data tweets for 5 months starting from September 1, 2019 to 31 January 1, 2019. Indonesia, namely Jakarta, Bandung, Medan, and Surabaya. Each data will be taken manually by using the Geolocation API that has been provided by Twitter via a Twitter search. The results of the classification using the Naïve Bayesian Classifier algorithm received 839 positive tweets, 32 negative tweets, and 67 neutral tweets from 938 overall tweets, or in the form of a percentage, there were 90% containing positive sentiment, 3% negative, and 7% negative sentiment towards Mr. Joko Widodo. And 56 positive tweets, 6 negative tweets, and 8 neutral tweets from 70 overall tweets, or in the form of the percentage there are 80% positive sentiments, 9% negative sentiments, and 11% neutral sentiments towards Mr. Prabowo. The level of accuracy generated from the Naïve Bayesian Classifier algorithm itself for this study amounted to 77.62%. Keywords: sentiment analysis, Indonesian presidential candidate pair, jokowi, prabowo, geolocation fire, naïve bayesian classifier, Multinomial Naive Bayes.