Abstract:To reveal the public perception of cultural ecosystem services (CES) in recreational spaces in Lanzhou, this study integrates natural language processing (NLP), driver-based importance–performance analysis (IPA), and semantic network analysis. Based on 31 350 online reviews, it examines the perceived characteristics of CES in Lanzhou's recreational spaces. The results show that public perceptions of CES are primarily concentrated in two dimensions: direct visual experience and cultural–educational cognition, while a mismatch exists between service supply and demand. Specifically, aesthetic experience receives relatively high satisfaction, whereas cultural and educational services exhibit a pattern of high importance but low satisfaction. This mismatch is mainly attributable to insufficient contextual interpretation of cultural resources, which keeps visitors' perceptions of historical relics, architectural landmarks, and other tangible resources at a superficial recognition level, making it difficult to move toward deeper meaning-making and cultural identification. In addition, basic visitation and management factors, such as ticketing and crowding, hinder the in-depth transmission and experience of cultural value to some extent. Accordingly, this study suggests that the optimization of recreational spaces should focus on narrative guidance for cultural settings and the improvement of basic service facilities, thereby facilitating the transformation of recreational resources into deeper cultural experiences.
