Analisis Sentimen Terhadap Kontroversi Putusan MK Mengenai Usia Capres-Cawapres Menggunakan Multi-Layer Perceptron Dengan Teknik SMOTE

Sasmita, Sasmita
Jariah S.Intam, Rezki Nurul (Universitas Negeri Makassar)
Surianto, Dewi Fatmarani (Universitas Negeri Makassar)
B, Muhammad Fajar (Universitas Negeri Makassar)

Faktor Exacta
Vol 17, No 2 (2024)

In October 2023, the Constitutional Court's decision on age limit requirements for presidential and vice-presidential candidates stirred controversy, perceived as favoring a specific vice-presidential candidate. Public reactions flooded social media platforms, particularly on Najwa Shihab's YouTube channel, where sentiment analysis was conducted on 505 comments under the video titled "Putusan MK: Publik memang Seharusnya Marah" (Constitutional Court Decision: The Public Should Indeed Be Angry). The comments were categorized into three sentiment classes: 425 negative, 42 neutral, and 38 positive. The study employed Multi-Layer Perceptron (MLP) models tested on both imbalanced and balanced data using the SMOTE oversampling technique. Two feature extraction methods, TF-IDF weighting and countvectorizer, were applied. Results showed that the combination of TF-IDF with balanced data yielded the most effective classification model, boasting a remarkable accuracy, precision, recall, and F1-score, each at 99%.