Exploring the Digital Pulse of Sentiment Analysis on Online Gambling in Indonesia using χ² Feature Selection

Online gambling has sparked widespread discussions in Indonesia, prompting concerns among policymakers, researchers, and the public. This study applies sentiment analysis to online discourse related to gambling, leveraging machine learning techniques and chi-square (χ²) feature selection to identify key sentiment-driving words. Public sentiment is categorized into positive, neutral, and negative classes, with χ² scores ranking significant terms that contribute to each category. The results reveal a predominant negative sentiment, with words such as stupid, disturbing, and dangerous frequently appearing in gambling-related discussions. Meanwhile, neutral terms like more and really exhibit high statistical relevance, suggesting their contextual ambiguity. These findings provide valuable insights for policymakers and digital platforms in shaping public awareness and regulatory measures. Future research can extend this approach using deep learning-based feature selection to refine sentiment classification and improve predictive accuracy.