An innovative approach to urban parks and perception: a cross-cultural analysis using big and small data

Abstract

Abstract Urban parks provide essential benefits that enhance human well-being and quality of life. This study integrates big data (online reviews, images) and small data (survey results) with large language models (LLMs) and object detection algorithms to analyze public perceptions of urban parks in Stockholm, New York, and Shanghai. LLMs were employed for sentiment analysis and keyword extraction from 62,724 online reviews, enabling the identification of key environmental features influencing visitor satisfaction. Concurrently, the YOLO v11 object detection model analyzed 111,469 images to quantify the presence of natural and built features, such as greenery, water bodies, and recreational facilities. Findings reveal that online satisfaction scores are lower than survey-based scores in Stockholm (0.661 vs. 0.774) and Shanghai (0.626 vs. 0.845), indicating a negativity bias in online reviews. In contrast, New York’s online (0.612) and survey (0.610) scores align closely, suggesting a broader representation of perspectives. Feature analysis shows greenery, flowers, and recreational facilities consistently enhance satisfaction, while noise and uncleanliness reduce it. Cultural differences emerge: small animals significantly improve perceptions in New York, whereas excessive liveliness lowers satisfaction in Shanghai due to crowding concerns. By examining 102 parks across these cities, this study highlights the effectiveness of AI-driven, multi-source data analysis in identifying universal and culturally specific park preferences. These insights can guide data-informed urban park management strategies, improving accessibility and visitor experiences globally.

Keywords

• Urban parks
• Public perception
• Large language models
• Cultural preferences
• Cross-cultural comparison