Sentiment Analysis of the Indonesia Presidential Election 2024 with Ensemble Learning for Reducing Disinformation on Social Media
Jihan Nabilah Hakim, Yuliant Sibaroni, Sri Suryani Prasetiyowati · 2023
Disinformation on social media is a frequent problem, especially in the political context related to the 2024 Presidential Election. The purpose of this sentiment analysis research is to reduce disinformation through the application of the Ensemble Learning method using 4 classifiers namely SVM, Random Forest, Gradient Boosting, and Stacked Ensemble, by utilizing TF-IDF and Cross-Validation. Data used in modeling is 11,611 data. The results of the research show that SVM, Random Forest, Gradient Boosting Classifier, and Stacked Ensemble are able to classify disinformation with significant accuracy. SVM achieved 86.44% accuracy, Random Forest achieved 87.65% accuracy, and Gradient Boosting Classifier achieved 88.20% accuracy. Stacked Ensemble achieves the same accuracy as GBC. Ensemble Learning through Stacked Ensemble has proven to provide promising results in efforts to reduce disinformation on social media. The results of this research make an important contribution in developing the use of sentiment analysis to address the problem of disinformation in social media