Sentiment Analysis Pipeline for Hospital Reviews in Yogyakarta

To transform how healthcare quality is monitored in Yogyakarta, Indonesia, this project introduces an end-to-end automated sentiment analysis pipeline that extracts actionable insights from over 22,000 patient reviews on Google Maps. The system leverages the Google Maps Places API and a custom Selenium scraper for data collection, feeding raw Indonesian text through a rigorous preprocessing phase before classifying sentiments using a fine-tuned IndoBERT model that achieves an impressive 82% accuracy. Beyond basic classification, the pipeline performs advanced analytics such as sentiment driver identification, anomaly detection, and rating gap analysis, revealing crucial service nuances that simple star ratings often obscure. All of these insights are managed through a PostgreSQL and FastAPI backend, culminating in a modern, interactive Next.js dashboard that empowers hospital administrators and policymakers with a scalable, containerized tool to continuously evaluate and improve patient satisfaction.