Analisis Sentimen Komentar YouTube pada Video Clash of Champions di Channel Ruangguru Menggunakan Algoritma BERT

The sentiment analysis technique aims to understand public opinions through textual data,including YouTube comments. These comments often reflect diverse perspectives from the audience regarding specific content. This research examines sentiment analysis of YouTube comments on the Clash of Champions video from Ruangguru’s channel by utilizing the Bidirectional Encoder Representations from Transformers (BERT) model. The primary goal of this study is to identify and analyze sentiments within these comments using the BERT model, renowned for its advanced capability in comprehending natural language. The dataset used comprises YouTube comments that have been pre-processed, manually labeled, and supplemented with data augmentation techniques to ensure the quality and reliability of the analysis. The implementation employs the indobenchmark/indobert-base-p1 library of the BERT model and evaluates its performance using metrics such as accuracy, precision, recall, and F1-score. The analysis demonstrates that the model successfully achieves an accuracy of up to 96% in categorizing sentiments into four groups: positive, negative, neutral, and others. These findings provide Ruangguru with valuable insights into public sentiment and offer strategic recommendations for enhancing the quality of its content.