Towards Intrinsic Quality Analysis of OpenStreetMap Datasets Christopher Barron , Pascal Neis, Alexander Zipf Geoinformatics Research Group, Institute of Geography Heidelberg University Christopher Barron, Geoinformatics Research Group, Institute of Geography, Heidelberg University
Mapping, Routing, LBS, etc. No intercalated instance for quality assurance! Christopher Barron, Geoinformatics Research Group, 2 Institute of Geography, Heidelberg University
Research on OSM Data Quality • Road Network: Haklay (2008), Ather (2009), Kounadi (2009), Ludwig (2010), Zielstra & Zipf (2010), Helbich et al. (2012), Neis et al. (2012), etc. • Polygon Features: Girres & Touya (2010), Mooney et al. (2010), etc. Reference Dataset Christopher Barron, Geoinformatics Research Group, 3 Institute of Geography, Heidelberg University
Problem Reference Dataset: • Availability • Licensing Restrictions • Costs • Actuality • Suitability Christopher Barron, Geoinformatics Research Group, 4 Institute of Geography, Heidelberg University
Motivation How can we evaluate OSM datasets without a reference dataset? Christopher Barron, Geoinformatics Research Group, 5 Institute of Geography, Heidelberg University
Christopher Barron, Geoinformatics Research Group, 6 Institute of Geography, Heidelberg University
Christopher Barron, Geoinformatics Research Group, 7 Institute of Geography, Heidelberg University
Methods for Quality Analysis based on an OSM-Full-History- Dump Intrinsic Approach Christopher Barron, Geoinformatics Research Group, 8 Institute of Geography, Heidelberg University
iOSMAnalyzer • Framework for intrinsic OSM quality analysis • Command line based tool • Output: statistics, maps & plots • ‘ Fitness for Purpose ’ Christopher Barron, Geoinformatics Research Group, 9 Institute of Geography, Heidelberg University
iOSMAnalyzer - Architecture Christopher Barron, Geoinformatics Research Group, 10 Institute of Geography, Heidelberg University
iOSMAnalyzer – Based on Open Source Components Christopher Barron, Geoinformatics Research Group, 11 Institute of Geography, Heidelberg University
iOSMAnalyzer - Framework General Area Information Address-Search • • Development of points, lines & polygons Increase/decrease of address information over time • • Actuality of the dataset Completeness of house numbers tagged to buildings • Comparison of newly created and edited objects Points of Interest- Search • Quantitative analysis in respect to the attributes of OSM features • Development of POIs & average number of • attributes Syntactic attribute accuracy • • Attributive completeness of several POI categories Positional accuracy of road intersections from different timestamps Map-Applications Routing & Navigation • Development of the geometry representation of • polygons Development of the total road network length • • Untouched points, lines & polygons Development of the syntactic attribute accuracy of roads • Development of invalid polygons • Actuality of the road network • Logical consistency of landuse polygons • • Logical consistency of the total road network Quantitative development of selected polygons • Roads without a name or route number User -Information & -behavior • Grid-based analysis of the positional accuracy of the • Number & activity of contributors total road network • Distribution of contributions • User profiles Christopher Barron, Geoinformatics Research Group, 12 Institute of Geography, Heidelberg University
iOSMAnalyzer – Example (1/2) General Area Information Address-Search • • Development of points, lines & polygons Increase/decrease of address information over time • • Actuality of the dataset Completeness of house numbers tagged to buildings • Comparison of newly created and edited objects Points of Interest- Search • Quantitative analysis in respect to the attributes of OSM features • Development of POIs & average number of • attributes Syntactic attribute accuracy • • Attributive completeness of several POI categories Positional accuracy of road intersections from different timestamps Map-Applications Routing & Navigation • Development of the geometry representation of • polygons Development of the total road network length • • Untouched points, lines & polygons Development of the syntactic attribute accuracy of roads • Development of invalid polygons • Actuality of the road network • Logical consistency of landuse polygons • • Logical consistency of the total road network Quantitative development of selected polygons • Roads without a name or route number User -Information & -behavior • Grid-based analysis of the positional accuracy of the • Number & activity of contributors total road network • Distribution of contributions • User profiles Christopher Barron, Geoinformatics Research Group, 13 Institute of Geography, Heidelberg University
Total Road Network Length (1/2) Assumption (Neis et al. 2012): 1) Motorways/Highways Mapped completely 2) Secondary/Tertiary Roads 3) Residential Roads 4) Tracks, Paths, etc. Christopher Barron, Geoinformatics Research Group, 14 Institute of Geography, Heidelberg University
Total Road Network Length (1/2) Christopher Barron, Geoinformatics Research Group, 15 Institute of Geography, Heidelberg University
Total Road Network Length (1/2) Christopher Barron, Geoinformatics Research Group, 16 Institute of Geography, Heidelberg University
iOSMAnalyzer – Example (2/2) General Area Information Address-Search • • Development of points, lines & polygons Increase/decrease of address information over time • • Actuality of the dataset Completeness of house numbers tagged to buildings • Comparison of newly created and edited objects Points of Interest- Search • Quantitative analysis in respect to the attributes of OSM features • Development of POIs & average number of • attributes Syntactic attribute accuracy • • Attributive completeness of several POI categories Positional accuracy of road intersections from different timestamps Map-Applications Routing & Navigation • Development of the geometry representation of • polygons Development of the total road network length • • Untouched points, lines & polygons Development of the syntactic attribute accuracy of roads • Development of invalid polygons • Actuality of the road network • Logical consistency of landuse polygons • • Logical consistency of the total road network Quantitative development of selected polygons • Roads without a name or route number User -Information & -behavior • Grid-based analysis of the positional accuracy of the • Number & activity of contributors total road network • Distribution of contributions • User profiles Christopher Barron, Geoinformatics Research Group, 17 Institute of Geography, Heidelberg University
Positional Accuracy of Junctions (2/2) Helbich et al. (2012) Christopher Barron, Geoinformatics Research Group, 18 Institute of Geography, Heidelberg University
Positional Accuracy of Junctions (2/2) Jan. 2012 & Jan. 2013 Christopher Barron, Geoinformatics Research Group, 19 Institute of Geography, Heidelberg University
Summary • Q = {V 1 , V 2 … V n-1 } • OSM quality analysis … – based on OSM-Full-History-Dumps; – are possible without reference datasets; – are arbitrarily repeatable. Christopher Barron, Geoinformatics Research Group, 20 Institute of Geography, Heidelberg University
Future Work • Include further methods & indicators • Consideration of relations • Contributor profiles • Vandalism detection • Analysis on different scales (e.g. country) • Performance improvement Christopher Barron, Geoinformatics Research Group, 21 Institute of Geography, Heidelberg University
Thank you for your attention! Christopher Barron Geoinformatics Research Group Institute of Geography Heidelberg University http://giscience.uni-hd.de christopher.barron@geog.uni-heidelberg.de Christopher Barron, Geoinformatics Research Group, 22 Institute of Geography, Heidelberg University
Recommend
More recommend
