Sentiment analysis of Indonesian presisential election 2019 on the twitter with lexicon-based and support vector machine (SVM)
Fredericus Dwi Nugraha Putra, Pranowo Pranowo, Djoko Budiyanto Setyohadi · AIP conference proceedings · 2020
The development of social media is currently very rapid, and Twitter is a social media that is widely used by people in disseminating information, even in the presidential election process of social media twitter has an important role in the dissemination of good or evil information to drop the image of one presidential candidate. This study aims to analyze all information obtained from Twitter to obtain positive and negative values to obtain the electability predictions of presidential candidates. In the process of classification, this research uses the lexicon-base method and support vector machine (SVM). The preprocessing data stage uses part-of-speech tagging, chi-square test, and aggregating opinion on the entity (NN & NNP) which aims to avoid low recall from the lexicon-based method. The data used takes tweets both in Indonesian and english with several datasets of 2000 tweets with the separation of two opinions from each presidential candidate. From the polarity accuracy, get the value with the jokowi keyword to get a positive value of 24.10%, negative 38.30% and neutral 38.30% while for the keyword prabowo it gets a positive value of 0.20%, negative 0.10% and neutral 0.70%.