Identifying the Experiential Qualities of Landscapes: an Exploration of Artificial Intelligence Techniques

This paper explores the use of on-site experiences and Natural Language Processing techniques for identifying meaningful experiential descriptors of preferred landscape settings. Three groups of participants (N=25) were taken on walks through a varied landscape and asked to rate their preferences and describe in their own words their feelings about each of 16 landscape settings. Three weeks later the same participants completed a similar exercise, but this time responding to slides of the landscape settings. Using a grammar-based natura/language processing program, results showed that on-site experiences provided a richer diversity and content of descriptors than could a laboratory experience. Artfficiallntelligence techniques such as this, and new advances in neural networking approaches as well, hold considerable promise for identifying the experiential qualities of landscapes.