Text Emotion Detection: Discover the Meaning Behind YouTube Comments Using Indo RoBERTa

The rapid growth of YouTube videos is directly proportional to their comments. This comment attracts researchers to extract the meaning of public response and feedback about specific content. Utilizing insights from advertising YouTube comments can be beneficial in evaluating advertising. To see if the company can increase the effectiveness of the ad with emotion detection. This study build deep learning text-based emotion detection uses CNN, Multinomial Naive Bayes, and Indo RoBERTa to compare and evaluate which one is the most accurate and uncover the emotions behind comments. The models are extract the emotion of Indonesian comments as any one of the five basic emotions (anger, fear, joy, sadness, and surprise). Relating to the reaction of the public to the advertisement for one of the shampoo products on the YouTube Channel, the result shows that the text’s emotional expressions are primarily those of joy and surprise. This study successfully detects five text emotions and identifies Indo RoBERTa as the best model for domain and language-specific text-emotion detection.