武夷山文献分析
李凤娇等
Source document (2026)
Based on a comprehensive review of literature related to the built environment of Wuyishan, this study proposes a large language model-based method for the intelligent identification of human settlement features. Through a systematic screening and analysis of Chinese and English publications on Wuyishan from 2015 to 2025, six core dimensions of the built environment were established: natural ecology and landscape, spatial structure and development intensity, housing and community, functional and service facilities, transportation and accessibility, and socio-cultural perception and experiential quality. Using the ChatGPT-4.1 model, built environment feature terms were automatically extracted from the literature and subjected to sentiment analysis. Furthermore, the results were examined through a dual comparison framework, integrating features derived from social media data and manually annotated references. The findings indicate that the large language model demonstrates high stability and accuracy in feature extraction and sentiment analysis. Natural ecology and landscape, as well as socio-cultural perception and experiential quality, emerge as the key strengths of Wuyishan's built environment, while functional and service facilities and transportation and accessibility are identified as priority areas for improvement. The housing and community dimension represents a medium- to long-term optimization direction. Overall, this research expands the application of large language models in urban and rural planning and provides theoretical support for spatial optimization, local industrial development, and ecological civilization construction.
