Within textual data there is a potential for obtaining spatial information. This can help to locate people based on their relative distance information, without requiring extensive knowledge of the environment. To test this heuristic approach, the data from the Airbnb online platform was used as the basis.The suitability of this source is sincethe relative location of the accommodation to tourist attractions is an important criterion when booking. Text mining and geoparsing techniques were used

more »

... to determine relevant triplets from the existing corpus. This ensured that only geo-text data remained for the analysis, that is, theoretically only information that had the potential to find positions based on quantitative distance information. It has been shown that of the 293 detected sentences, 20% could be transferred as relevant distance information into a geographic information system (GIS). It was based on a manual coding, as in the sentences semantic ambiguities occurred and the relatum was often not identified by the computer. The subsequent analysis showed that with two relative distance indications, no further processing is necessary, thereby reducing the distance to the potential area in which the Airbnb is located to 13.8m. In the presence of more than two indications, a visual interpretation had to be applied because the uncertainties were too great. There are several overlapping areas, some of which cannot be automatically reduced. Nevertheless, in these cases, with additional consideration of the real use of the buildings, even the relevant block of houses can be determined.