Analisis Sentimen terhadap Terorisme pada Platform Twitter menggunakan Support Vector Machine

This research aims to classify public sentiment regarding terrorism issues using the Support Vector Machine (SVM) algorithm. This topic is important because text-based sentiment analysis plays a significant role in understanding public opinion on critical issues. Initial data in the form of Indonesian text was processed through preprocessing stages, translated into English, and labeled using VADER. Data imbalance was addressed using Random Over Sampling methods, while numerical data representation was obtained through feature extraction using TF-IDF. The SVM model was evaluated using confusion matrix with accuracy, precision, recall, and F1-score metrics. The results show that the model achieved 98.02% accuracy, 98.09% precision, 98.02% recall, and 98.01% f1-score, demonstrating excellent performance in classifying sentiment into negative, neutral, and positive categories. Some prediction errors were still found in the negative and positive categories. This research demonstrates that the combination of preprocessing methods, data balancing, and TF-IDF feature extraction effectively produces an accurate sentiment classification model. This research contributes significantly to the development of text-based sentiment analysis technology to support decision making. Keywords: Sentiment Analysis, Support Vector Machine, Terrorism, Twitter Penelitian ini bertujuan mengelompokkan sentimen masyarakat terkait isu terorisme menggunakan algoritma Support Vector Machine (SVM). Topik ini penting karena analisis sentimen berbasis teks berperan signifikan dalam memahami opini publik terhadap isu-isu kritis. Data awal berupa teks berbahasa Indonesia diproses melalui tahap preprocessing, diterjemahkan ke bahasa Inggris, dan dilabeli menggunakan VADER. Ketidakseimbangan data diatasi dengan metode Random Over Sampling, sementara representasi data numerik diperoleh melalui ekstraksi fitur TF-IDF. Model SVM dievaluasi menggunakan confusion matrix dengan metrik accuracy, precision, recall, dan f1-score. Hasilnya, model mencapai akurasi 98,02%, precision 98,09%, recall 98,02%, dan F1-score 98,01%, menunjukkan performa sangat baik dalam mengklasifikasikan sentimen ke dalam kategori negatif, netral, dan positif. Beberapa kesalahan prediksi masih ditemukan pada kategori negatif dan positif. Penelitian ini menunjukkan bahwa kombinasi metode preprocessing, penyeimbangan data, dan ekstraksi fitur TF-IDF efektif menghasilkan model klasifikasi sentimen yang akurat. Penelitian ini berkontribusi secara signifikan terhadap pengembangan teknologi analisis sentimen berbasis teks untuk mendukung pengambilan keputusan. Kata kunci: Analisis Sentimen, Support Vector Machine, Terorisme, Twitter