Using textual volunteered geographic information to model nature-based activities: A case study from Aotearoa New Zealand

Abstract

A boom in volunteered geographic information has led to extensive data-driven exploration and modeling of places. While many studies have used such data to explorehuman-environment interaction in urban settings, few have investigated natural,non-urban settings. To address this gap, this study systematically explores the contentof online reviews of nature-based recreation activities, and develops a fine-grained hierarchicalmodel that includes 28 aspects grouped into three main domains: activity, settings,and emotions/cognition. It further demonstrates how the model can be used to explore thevariation in recreation experiences across activities, setting the stage for the analysis of thespatio-temporal variations in recreation experiences in the future. Importantly, the studyprovides an annotated corpus that can be used as a training dataset for developing methodsto automatically capture aspects of recreation experiences in texts