Purpose: This study aims to analyze public sentiment regarding educational services in modern Islamic boarding schools using the Naïve Bayes method. The findings provide recommendations for improving educational quality. Methods: The research follows the Cross-Industry Standard Process for Data Mining (CRISP-DM) framework, utilizing web scraping techniques to collect data from social media and online discussion forums. The Naïve Bayes algorithm is used for sentiment classification. Result: A dataset of 387 reviews was analyzed, showing that 82.8% of reviews were positive, while 17.2% were negative. The model achieved an accuracy of 88%. Novelty: Unlike previous studies, this research focuses specifically on modern Islamic boarding schools, employing machine learning for sentiment classification to provide actionable recommendations.