An assessment of park satisfaction and its associated factors in Zhengzhou

Traditional urban park satisfaction assessments commonly adopt static, single-source datasets, which are insufficient to capture dynamic shifts under emergency contexts. This study proposes a multi-source evaluation framework and applies it to 19 parks in Zhengzhou, China, leveraging social media data (2020–2025) and advanced artificial intelligence models (BLIP-2, BERTopic, GPT-4o) for analysis. The results reveal divergent satisfaction trajectories across park types: comprehensive parks maintain relatively high and stable satisfaction, community parks show an overall post-pandemic upward trend in satisfaction, whereas specialty parks generally record lower satisfaction levels. Further analysis identifies distinct associated indicators for each park category, with the number of activity items associated with comprehensive parks, infrastructure completeness associated with community parks, and plant species associated with specialty parks. This study explores temporal shifts in park satisfaction via emotional expression analysis and characterizes associations between associated variables. Building on these findings, targeted strategy recommendations are proposed to improve satisfaction tailored to each park type.