Creative Commons License
This work is licensed under a Creative Commons Attribution-Share Alike 4.0 License.
The user experience of any OpenStreetMap (OSM) based service heavily depends on the quality of the underlying data. If the service deals with points-of-interest (POIs), consistent and comprehensive tagging of the respective map elements is a necessary condition for a satisfying service. In this paper, we develop methods that can automatically infer tags characterizing POIs solely based on the POI names. The idea being that many POI names already contain sufficient information for tagging. For example, 'Pizzeria Bella Italia' most certainly indicates an Italian restaurant. As the OSM data contains hundred of thousands POIs for Germany alone, we aim for a tool that can accomplish tag extrapolation in an automated way. In a first step, we automatically extract typical words and phrases that occur in names associated with a certain tag. For example, learning indicators for ‘shop=hairdresser’ on German OSM tags led to high scores for ‘fris’, ‘cut’, hair’ and ‘haar’. Having available such indicator phrases, we use standard machine learning techniques to derive the probability for a POI to exhibit a certain tag. If this probability exceeds a certain threshold, we assign the tag to the POI in an automated fashion. We used our extrapolation framework to create new amenity, shop, tourism, and leisure tags. The accuracy of our approach was over 85% for all considered tags. Moreover, for POIs tagged with amenity=restaurant, we aimed for extrapolating the respective cuisine tag. For more than 19 thousand out of 28 thousand restaurants in Germany lacking the cuisine-tag, our approach assigned a cuisine. In a random sample of those assignments 98% of these appeared to be true.
Storandt, Sabine and Funke, Stefan
"Automatic Improvement of Point-of-Interest Tags For OpenStreetMap Data,"
Free and Open Source Software for Geospatial (FOSS4G) Conference Proceedings: Vol. 15
, Article 56.
Available at: https://scholarworks.umass.edu/foss4g/vol15/iss1/56