Spatial sentiment analysis of tourist points of interest

Thanks to the Web, tourists’ experiences and opinions are more and more valued resources for creating a new symbolic identity for an attraction or a place, where tourists want to live a unique immersive experience rather than being only temporary bystanders. Textual reviews of points of interest (attractions, restaurants, accommodations) on platforms such as TripAdvisor can be processed and used in statistical analyses to obtain useful insights. Here we propose to compute the reviews' polarity scores and use them in combination with other characteristics (e.g., price, offered services and type of tourist facilities) to build spatial clusters of tourist points of interest. The geo-referenced semantic orientation of reviews concerning a particular activity or attraction represents a useful quantitative feature for further analyses and for the production of territorial statistics. We aim to use geo-referenced polarity scores to quantitatively and spatially assess whether and how tourist sentiment towards a place changed. Our strategy is developed under an Ambient Geographic Information (AGI) framework, in which social media are used to understand human narrative experiences and their evolution over time. The proposal can be extended to monitor the change of sentiment towards specific areas of interest and plan possible intervention policies.