Sentiment Analysis of Quick Count Results of the 2024 Indonesian Presidential Election on Social Media

Sentiment analysis is a technique of analyzing text to determine the emotion of the message. The main objective of this sentiment analysis is to investigate whether Indonesian citizens on the platform X felt positive, negative, or neutral towards the Quick Count results of the recent 2024 Indonesian Presidential Election, in light of public distrust surrounding the legitimacy of the election organizers. The goal of this study is also to compare the performance of two classifier models; Naive Bayes and LSTM, based on standard metrics like accuracy, precision, recall, and f1-score. The data used were posts scraped from X taken after the election day on February 14th, along with datasets sourced from Kaggle and GitHub. The data preprocessing followed a sequence of data cleansing, case folding, and stop word removal. The final dataset is then split and vectorized using TF-IDF for Naive Bayes, and tokenized for LSTM. The result of this research shows 75.142% of accuracy for the Naive Bayes classifier and 80.03 % of accuracy for the LSTM-based classifier. From the 2,639 posts classified by the Naive Bayes model, 64.7% show positive sentiment towards the Quick Count results, whereas the LSTM model produced 58.2% positive sentiment of the same amount.