Crowdsourced text analysis to characterize the U.S. National Parks based on cultural ecosystem services

Abstract

Perceiving cultural ecosystem services (CES) are subject to physical landscape settings and sociocultural context. However, we still have a limited understanding of multiple CES and how they differ across heterogeneous landscapes. This study aims to identify and compare eight different CES across 48 National Parks in the United States. The study uses TripAdvisor reviews as a source of data and applies text analysis and a crowdsourced lexicon to assign CES for phrasal expressions.

We found that the parks can be grouped into four categories based on the composition of CES: (a) Aesthetic, (b) Biological/Spiritual, (c) Recreation/Identity, or (d) Cultural/Educational/Social values. The most commonly used phrases in the reviews correspond to the specific CES group of the park, while describing key attractions and popular activities. The geographic distribution of the parks also revealed spatial autocorrelations depending on the CES group. Overall, the study demonstrates how crowdsourced text analysis can be used to inform data-driven management strategies that promote unique niches for tourism while diversifying conservation agendas.