Abstract

Urban Green Spaces (UGS) play a crucial role in improving environmental quality, fostering social interaction, and enhancing the well-being of urban communities. In Magelang City, Indonesia, a growing small city, understanding public perceptions of UGS is essential to ensure their effective development and to align with Sustainable Development Goal (SDG) 11: Sustainable Cities and Communities. This study employs an AI-based approach to evaluate public sentiments and identify key discussion themes by applying big data analytics to user-generated reviews from Google Maps. Python-based text mining techniques were utilized, with the Valence Aware Dictionary for Sentiment Reasoning (VADER) used for sentiment classification and Latent Dirichlet Allocation (LDA) applied for topic modeling. Results show that public perception is predominantly positive, emphasizing cleanliness, comfort, aesthetic value, and accessibility. Negative sentiments, although fewer, highlight issues in facility maintenance, limited amenities, safety, and spatial accessibility. These findings provide actionable implications for policymakers by offering evidence-based justification for future investments, responsive design strategies, and continuous monitoring of UGS quality from a citizen-centered perspective. Contributing to more inclusive, safe, and sustainable urban development aligned with SDG 11.7.

Keywords: Urban Green Spaces, Sentiment Analysis, Topic Modeling, Text Mining, Sustainable Urban Development.