YouTube Comment Sentiment Analysis on Deddy Corbuzier and BEM UI’s Podcast Using TF-IDF and Naïve Bayes

Abstract

The development of technology and the popularity of digital platforms such as Youtube have had a significant impact on the dissemination of information, especially related to the idea of a presidential candidate in the 2024 election. For example, Deddy Corbuzier actively uploads a podcast video on his Youtube channel, together with the chairman of BEM UI, to discuss the ideas of each presidential candidate. In this study, sentiment analysis was carried out on more than 2.6 million Youtube user comments on the video using the Naïve Bayes Classifier algorithm. This algorithm has proven effective in previous studies, showing high accuracy in classifying people's sentiments. The research methodology includes data labeling, text preprocessing, word weighting with TF-IDF, data validation using k-fold cross validation, and data testing. The results of the sentiment analysis revealed that more than fifty percent of the comments were positive, while some remained neutral. The data visualization process using word cloud provides a clear picture of the topics most talked about by the public, with the word "Leader" dominating. Although the accuracy of the Naïve Bayes Classifier in this study reached 57.6%, this study provides valuable insights into the public's view of the idea of a presidential candidate. Further development may involve the use of other algorithms as comparators to improve the accuracy of sentiment analysis.