Sentiment Analysis of Indonesian TikTok Comments Using TF‑IDF with Naive Bayes and SVM

Abstract

This study aims to develop an automatic sentiment classification model for Indonesian TikTok comments using Term Frequency–Inverse Document Frequency (TF‑IDF) with Naive Bayes and Support Vector Machine (SVM). Fifteen thousand comments were collected from public TikTok videos and manually labeled as positive, negative, and neutral. Data preprocessing included case folding, tokenization, stopword removal, and stemming (Nazief‑Adriani algorithm). TF‑IDF weighting transformed text into vectors, then used to train both classifiers. Performance was evaluated using accuracy, precision, recall, and F1-score trough 5-fold cross-validation. Result show SVM outperforms Naive Bayes with 92.8% accuracy versus 83%. Findings confirm that TF-IDF combined with SVM produces more relieble result for short Indonesian text classification, offering valuable insights for social media monitoring applications.