The policy of shutting down TikTok Shop has sparked both pros and cons. On one side, it eliminates jobs for content creators whose income relies on TikTok Shop, while on the other side, it saves UMKM from predatory pricing wars that harm them. Utilizing the Naive Bayes algorithm, a classification method capable of predicting the likelihood of a class and making decisions based on learning data, the Emotion Recognition research on YouTube comments related to the closure of TikTok Shop is conducted. Data will be classified into five classes: happy, angry, sad, afraid, and surprised. The objective of this research is to find the best emotional model using the Naive Bayes method. The results of user testing with Naive Bayes and Tf-Idf show that the precision values for sad, happy, afraid, and surprised emotions are high, while for anger, the percentage is 59%. The percentages for afraid, happy, sad, and surprised emotions are 91%, 87%, 84%, and 79%, respectively. The overall accuracy is 82%.
