Quantifying cultural ecosystem services in urban parks using social media and large language models: Insights for Beijing’s Garden City initiative

Urban parks are vital components of sustainable cities, providing cultural ecosystem services (CES) that enhance residents’ well-being. Yet understanding how people perceive these intangible benefits remains challenging. This study refines existing CES research by integrating large-scale social-media analytics with large language models (LLMs) to capture and interpret public perceptions of parks in Beijing’s core districts. Drawing on more than 46,000 user-generated comments, the LLM-based framework identifies six CES categories (i.e., historical & cultural, aesthetic, recreational, social-interaction, educational, and spiritual values) and evaluates their sentiment tendencies. The results reveal that aesthetic value is the most positively perceived dimension, while recreational, educational, and social-interaction values show moderate emotional intensity, indicating potential areas for experiential enhancement. Historical & cultural and spiritual values, though less frequently mentioned, generate strong positive affect when recognised. Integrating these findings through Importance-Performance Analysis highlights clear renewal priorities for Beijing’s Garden City initiative. Methodologically, the study demonstrates how LLMs can classify Chinese-language social-media text with high accuracy (F1 = 0.8945) and minimal manual annotation, offering a scalable, interpretable, and cost-effective approach for CES assessment. The combined use of social-media data, semantic modelling, and performance analysis provides transferable evidence for guiding park management and renewal in other high-density urban contexts. • Applies large language models for CES classification without additional training data • Integrates multi-source social media analytics with sentiment and IPA evaluation • Identifies aesthetic value as a key strength, with recreation, education, and social interaction as priorities for improvement • Demonstrates feasibility of LLM-based CES assessment for urban park renewal • Provides evidence-based insights for park renewal to support Beijing’s Garden City initiative