

Social media analysis has become a major instrument for data-driven tourism. It allows surveying visitor behavior on multiple scales. Considering the geographical characteristics of users’ posts from social media platforms, we were able to address more specific questions related to the place type selection patterns of the visitors. In this paper, we present OPENLOSTCAT, our first-order-logic-based location categorizer applicable for modeling location types depending on Open-StreetMap data. We report our findings revealed by this tool on more than one year’s collection of global-scale geotagged Twitter data, focused on potential trail-related hiking and trekking activities. We categorized visited locations in our experiments based on place accessibility – transport and trail infrastructure –, and analyzed these categories according to the travel distance taken by visitors in general to reach these areas. Our comparisons reveal seasonal characteristics, continental differences (between Europe and North America), as well as specifics related to selected recreational areas. Besides these preliminary findings available for further verification, we show both the perspectives and limitations of our approach for future improvements and experiments.