From text to insights: leveraging NLP to assess how landscape features shape tourist perceptions and emotions toward traditional villages

Abstract

Understanding tourist perceptions and the relationship between landscape features and emotional attitudes in traditional village is crucial for sustainable development. However, quantifying these perceptions over vast spatiotemporal scales is challenging. Additionally, there is a paucity of knowledge on which landscape features influence tourist perceptions within varied samples on a spatial scale. In this study, we analyzed 39,130 online reviews of 57 traditional Chinese villages from 2018 to 2023. Utilizing Word2Vec and machine learning techniques, we identified 14 key landscape feature indicators. These were quantitatively scored using expert evaluations, and their relationship to tourists’ emotional attitudes was determined through linear regression analysis. The analysis revealed that architectural elements, service facilities, convenience, and sanitation are positively associated with tourists’ emotional attitudes towards traditional villages. Contrary to expectations, the frequency of landscape feature mentions did not correlate significantly with emotional attitudes, challenging traditional assumptions about the visibility of landscape features and their impact.

Keywords

• landscape perception
• traditional villages
• online text
• natural language processing (NLP)
• machine-learning