<![CDATA[The rapid growth of social media has transformed digital communication, enabling citizens to express opinions openly on various public issues. This study examines YouTube user comments on the video "Cerdas Cermat Empat Pilar MPR RI West Kalimantan Province" to understand public sentiment toward civic education content. Manual analysis of large-scale YouTube comment data is inefficient and prone to subjectivity. This study addresses the question: how can Lexicon-Based Sentiment Analysis effectively classify public opinion in Indonesian-language YouTube comments? This study applies a lexicon-based rule-based approach without training data requirements, providing an accessible alternative for sentiment analysis on medium-scale Indonesian-language datasets with full preprocessing pipeline integration. A total of 1,709 comments were collected via web scraping using the YouTube Comment Downloader library on Google Colaboratory. Data preprocessing included case folding, text cleaning, stopword removal, Sastrawi stemming, and negation handling. Sentiment scoring was performed using a customized Indonesian sentiment lexicon. Neutral sentiment dominated with 857 comments (50.15%), followed by negative sentiment at 516 comments (30.19%), and positive sentiment at 336 comments (19.66%). Classification errors were identified in negation and sarcasm processing. Lexicon-Based Sentiment Analysis is a practical and efficient method for public opinion analysis on YouTube. The dominance of neutral comments indicates informative discussion patterns, while high negative proportions reflect a tendency toward public criticism. Future work should integrate machine learning comparisons and expanded lexicon resources.]]>
Copyrights © 2026
