Purpose: This study explores how user-generated content (UGC), specifically Google Maps reviews, can be used to assess public perception of urban green spaces (UGS) in six parks across Metro Manila and Seoul. It examines how natural language processing (NLP) tools can support park management by identifying areas for improvement. Method: Approximately 28,000 Google Maps reviews from six urban parks, Rizal Park, Quezon Memorial Circle, and Ayala Triangle Gardens in Metro Manila, and Seoul Forest, Boramae Park, and Yeouido Park in Seoul, published from 2015 to 2024 were collected using Apify’s Google Maps Reviews Scraper. The data was processed in Dataiku DSS and analyzed using Google Cloud NLP for multilingual sentiment classification and OpenAI’s GPT-4o Mini for multi-label thematic categorization. TF-IDF keyword extraction, combined with large language model (LLM) pre-processing, was applied to reviews tagged as “Points for Improvement and Negative Perceptions” to focus on issues park users encountered. Result: While most reviews were positive, many expressed neutral sentiment despite high star ratings, highlighting a disconnect between linguistic tone and numeric scores. Although fewer in number, negative reviews provided actionable insights into common issues such as overcrowding, cleanliness, accessibility, parking, and construction. Thematic analysis showed that in the Philippine parks, the most mentioned themes were general appreciation, history and culture, and recreation. In Korean parks, the focus was on recreational activities, general appreciation, and nature and biodiversity. The study demonstrates how UGC, combined with AI tools, can offer insights into urban park experiences, supporting more responsive and inclusive green space planning.
Espiritu et al. (Thu,) studied this question.