Scrutinizing the cultural ecosystem services of Chinese Classical Gardens: A novel deep learning approach based on online reviews from a multisensory perspective

The cultural ecosystem services (CESs) -related studies for decades have centered on the visual connotations of human perceptions rather than incorporating the other sensory experiences. This paper works to narrow the research gaps through scrutinizing the CESs of Chinese Classical Gardens from a multisensory perspective. We first demonstrate a qualitative phenomenological approach to conceptualizing the CES typologies of Chinese Classical Gardens, and then propose a novel deep learning approach to measuring their CESs based on online reviews from the lens of five senses. Following, the inter-relationships among the CESs are examined using co-occurrence network analysis. Results show that the CESs typologies of Chinese Classical Gardens include 7 main categories and 21 minor classes. Among them, the visual perception based CESs make up the highest proportion, but a noteworthy proportion of hearing, touch and taste based CESs is also observed. Additionally, the visual perception based CESs and auditory perception based CESs generally present higher centrality within the networked typology. Based on the discoveries, we finally discuss implications for landscape management. This paper foregrounds the effectiveness and feasibility of scrutinizing CESs from a multisensory perspective, and adds fuels to unpack the full spectrum of CESs for the geographical community.