On Tuesday 13 October, we have been more than happy to host volunteers of SAP in the course of the “SAP month of service” as our guests. They have supported us to merge volunteered data about accessibility from wheelmap.org and expert data about accessibility from Heidelberg Hürdenlos (a local accessibility platform). Heidelberg Hürdenlos contains more than 1,000 places rated by experts regarding their accessibility (e.g. “Alte Universität”).
Automation of such a data merging task can be more than tricky, due to different names or missing addresses of the Points of Interest (POIs) in the underlying datasets. However, for people with local knowledge, it is usually easy to decide which points belong to each other. In order to support this task, Sozialhelden have developed POIChecker.de which has been used by the volunteers.
During the half day event we have managed to match 516 of 1,026 POIs - 50%, a great result! 110 POIs got new accessibility attributes, 190 POIs have been confirmed, 26 have been upgraded to a better accessibility category (yes or limited) and, unfortunately due to stricter standards of the expert data, 190 POIs have been downgraded to a worse accessibility category (no or limited).
Since the volunteers had different knowledge and experience with POIChecker and OpenStreetMap before the event, some of the edits have introduced errors to the OSM, which is a normal observation for such volunteering mapping events. We will check and correct these errors within the upcoming weeks and we will also use this experience to further improve POIChecker.
Thanks to Timo and Adina from Sozialhelden and our student assistent Melanie Eckle for supporting this event, to our CAP4ACCESS colleagues from Fraunhofer IAIS for providing the change statistics and the cluster map, and of course and most importantly, thanks to all the volunteers from SAP!
If you click on the maps, you can view the changes on the Wheelmap before and after this event.
Detailed change statistics regarding the values of the OSM wheelchair Tag (prior/post event):