Quality Metrics for OSM:Directions & POI
Saturday, 10:05am – 20 min
In this talk, I will discuss research (as a part of Berkeley Undergraduate Research Program) that I have been doing along with Prof. Abhishek Nagaraj at UC Berkeley aiming to evaluate the quality of OpenStreetMap data. In particular, we are focused on understanding the relationship between OSM and well-established third party datasets in terms of (a) directions information and (b) Points of Interest. For directions, we have conducted research comparing over 100,000 different address pairs and differences in directions provided by OSRM as compared to Google Maps Directions API. Similarly, for POIs, we have programmatically compared a yellow-pages based third-party database with OSM data.
Our research up to this date is quite developed. We are finding that OSM compares quite favorably with Google Maps API in terms of directions and even outperforms Google in some cases. However, OSM’s POI completeness seems to be quite limited. We highlight areas where OSM is quite good and complete and where it needs improvements.
At SOTM, we hope to share these results with the community and get its feedback and guidance. This will be very valuable for our research moving forward.