Understanding the landscape of a modern Chinese city and summer resort from a missionary’s perspective: text mining ‘Beard Family Papers’ via large language models

AI-based text mining can be used to analyse the unstructured text of personal memories but has seldom been used in landscape research. Taking the example of Willard Livingston Beard, an important American missionary in Fuzhou, China, in the 19th and early 20th centuries, this research employs large language model (LLM) technology, including named entity recognition (NER), semantic and sentiment analysis (SA), and topic recognition (TR), to trace his footsteps through family letters and the sentiments expressed in his landscape descriptions. The results clearly reveal how the landscape of the summer resort served this missionary as a gazing object, social space, and ‘recharging station’, providing spiritual renewal. This research proves that, in the interpretation and explanation of landscape descriptions, AI tools significantly outperform traditional desk-based tools in terms of efficiency and consistency but rely on model training and validity testing. The results provide innovative referecess data for understanding the meaning of landscapes.