TY - JOUR AU - Siswanto Siswanto AU - Zakiyah Mar'ah AU - Alfiyah Sabir AU - Taufik Hidayat AU - Fadilah Adhel AU - Waode Amni PY - 2022/11/13 Y2 - 2025/04/03 TI - The Sentiment Analysis Using Naïve Bayes with Lexicon-Based Feature on TikTok Application JF - Jurnal Varian JA - Varian VL - 6 IS - 1 SE - Articles DO - https://doi.org/10.30812/varian.v6i1.2205 UR - https://journal.universitasbumigora.ac.id/index.php/Varian/article/view/2205 AB - On TikTok application, there are several types of content in the form of education, cooking recipes, comedy, various tips, beauty, business, etc. However, some non-educational contents sometimes appear on TikTok homepage even though minors can access the app. As a result, TikTok application can influence the behavior of minors to be disgraceful, therefore, an assessment of the application can be one of the objects for conducting sentiment analysis. The purpose of this study is to compare the results of sentiment analysis on TikTok application using Naïve Bayes with Lexicon-Based and without Lexicon-Based features. We used the TikTok reviews on Google Play Store as our data. According to the analysis, without Lexicon-Based feature, we obtained the accuracy rate, precision rate, and recall rate of 83%, 78%, and 69%, respectively. Meanwhile, the accuracy, precision, and recall rates using the Lexicon-Based feature were 85%, 91%, and 93%, respectively. Therefore, we concluded that sentiment analysis using Naïve Bayes with Lexicon-Based feature was better than without Lexicon-Based feature on TikTok reviews. ER -