ANALISIS SENTIMEN MASYARAKAT INDONESIA TERHADAP KASUS PERSETERUAN YAI MIM DAN SAHARA DI TIKTOK MENGGUNAKAN ALGORITMA NAIVE BAYES

Abstract

This study aims to analyze public sentiment toward the conflict case between Yai Mim and Sahara, which went viral on the TikTok platform. The data used in this study were TikTok user comments collected using Apify Instant Data Scraper, with a total of 10,504 comments. The research stages included data preprocessing (cleaning, normalization, tokenization, stopword removal, and stemming), sentiment labeling using a lexicon-based approach, feature weighting using the Term Frequency–Inverse Document Frequency (TF-IDF) method, and classification using the Naïve Bayes algorithm. As a comparison model, this study also implemented the Neural Network algorithm to compare classification performance. Model testing was conducted using four data split scenarios: 90:10, 80:20, 70:30, and 60:40 for training and testing data. The results showed that the Naïve Bayes model achieved the highest accuracy of 94.85% in the 90:10 scenario. Meanwhile, the Neural Network model demonstrated better performance with the highest accuracy of 96.49% in the 80:20 scenario. Based on these results, the 80:20 scenario was selected as the main reference because it provides a better balance in model evaluation. Overall, the combination of TF-IDF, Naïve Bayes, and Neural Network methods proved effective in classifying Indonesian sentiment comments on TikTok social media, with Neural Network showing more optimal performance compared to Naïve Bayes.