Two's a Crowd: Jerry Clough @ Open Addresses Symposium
1 like617 views
The document summarizes the experience of crowdsourcing addresses in OpenStreetMap in the UK. Primary data collection is done through bicycle or foot surveys of 250-1000 addresses per hour. Address data in OSM provides the largest openly licensed geolocated address dataset for the UK, though coverage remains patchy. While a small dedicated group contributes address data, growing the community and number of contributors remains a challenge. Various tools have been created to help with address mapping and quality assurance.
Two's a Crowd: Jerry Clough @ Open Addresses Symposium
- 1. Two's a Crowd:Two's a Crowd: OpenStreetMap Experience of Crowdsourcing Addresses in the UK Jerry CloughJerry Clough SK53 SK53.osm@gmail.com @SK53onOSM Blog: Maps Matter
- 3. SurveysSurveys ● Primary means of address collection – Bicycle or Foot – 250-1000 addresses per hour (inner city-suburbs) – Usually pencil & paper, perhaps with Field Papers. ● Data entry usually 2-3 times as long ● Experience dramatically improves efficiency ● Low immediate reward from mapping – Relatively few dedicated address mappers (~ 5% of active mappers) GPS traces of address surveys in Nottingham & Maidenhead. Source: OSM contributors & Mapbox
- 4. ResourcesResources ● Ground Surveys ● Local Knowledge ● Aerial Images – Bing / MapBox ● Licensed solely for OSM Mapping ● Photos – During survey – Geograph – Mapillary – OpenStreetView ● National Open Data – Ordnance Survey StreetView – LR Prices Paid, NROSH, Companies House ● Local Open Data – Planning – Food Hygiene ● Old Maps – NLS – OpenStreetMap Out-of- copyright maps
- 5. Typical OdditiesTypical Oddities ● Addresses not located on road different from name – Sherwin Walk, Nott'm ● Roads with houses, but no valid addresses – Sherwin Walk, Nott'm ● Roads with different names on each side – Poultry/Cheapside, Nott'm – Long Row/Smithy Row, Nott'm ● Multiple inconsistent addresses – Austin Reed, N'ham – (OSGB, NCC, Royal Mail & Austin Reed each has different version) ● More nesting of levels than supported by BS7666 – Leen Court, Nott'm – Named terraces (Bangor, N. Wales) Photo: 1-6, The Garland, Leen Court, Leen Gate, Nottingham NG7 2HR
- 6. DifficultiesDifficulties ● Gated Developments – Sheltered Housing – Modern flats & houses – Many modern social housing schemes ● Tower Blocks ● Radburn Estates ● In-fill of Victorian/Edwardian streets ● Absence of housenumbers ● Properties with names only
- 7. DataData
- 8. OSM Addresses in UKOSM Addresses in UK ● Largest OPEN set of accurately located data – Mostly to within 5 m – Some at delivery point ● Licensed under ODbL – Viral CC-BY-SA – Derived data an issue ● OpenData analogue of OSGB-derived data ● Created for: – Geolocation – Thematic interests e.g., MESH Edinburgh Historical Atlas (Richard Rodger) – Local Maps – General obsessiveness
- 9. Distribution of Address DataDistribution of Address Data ● > 800k (August '14) ● Places – Cambridge – Tendring – Wokingham – Birmingham – Nottingham – Broxtowe – Runcorn Excludes interpolated address (~ 100k) Areas shaded in deciles Centroid area scaled by number (Birmingham ~ 100k)
- 10. Patchy atPatchy at all levelsall levels
- 11. ToolsTools
- 12. KeyPad MapperKeyPad Mapper ● Dedicated Address Mapping – OpenSource Android app – Supported by ENIaKOON ● German telematics firm ● Owners active OSM contributors ● Simple to use – Data collected in OSM XML format – Location not accurate enough ● Smartphone GPS ● Canyon effect – Battery Life ● Other Smartphone apps – OSMAnd – Vespucci
- 13. Nominatim:Nominatim: Gazetteer, Geolocation, Geocoding ● Worldwide search engine ● Builds locational hierarchies ● Can use non-OSM Open Data – ONS CodePoint Open ● Many search strategies ● (Reverse) Geocoding
- 14. OSM-NottinghamOSM-Nottingham Integrates searching of OSM & Open Data ● Browsing of open data on a map ● Powerful tool to aid address / post code mapping ● Other – Restricted to some NG postcode districts – Thematic display of OSM data – Multiple raster layers http://osm-nottingham.org.uk/
- 15. Honourable MentionsHonourable Mentions ● PostCode FinderPostCode Finder – by Matt Williams ● No recent work (PhD to finish) ● Presentation at SotM13 – http://milliams.dev.openstreetmap.org/postcodefinder/ ● OsmoseOsmose – QA tool by Frédéric Rodrigo – Extensive used by OSM-FR and elsewhere ● MapRouletteMapRoulette – Platform for gamification of OSM edits – Martijn van Exel & Serge Wroclawski ● NYPL Building InspectorNYPL Building Inspector – Crowd-sourcing of parcels & addresses from historical NYC insurance maps – Developed by Tim Waters & Topomancy LLP ● And many others – Search & Geocoding: OpenCageData, MapZen – QA tools – Addressing tools: many by Svimik for Estonia
- 17. Motivating ContributorsMotivating Contributors More contributionsMore contributions ● Current rely on small core of address 'addicts' ● Steady drip of one-of contributions ● Germany has broader coverage – 10-times more OSMers – More activity in small towns – Local pride – More manageable task Imports of dataImports of data ● Imported data gets stale ● Quality rarely as good as surveyed data ● British community generally sceptical of value of imports ● Believed to have limited OSM in US
- 18. Project initiated by OpenStreetMap France BANO content : - OSM : 2.2 M addresses - opendata : 1.2M - cadastre : 14.9M As of August 2014 Green : OSM data Yellow : opendata Blue : cadastre + OSM* Red : cadastre only * matching roads/streets found in OSM data. 80 % of municipalities have a vector based cadastre http://openstreetmap.fr/bano BANOBANO
- 19. addresses.ioaddresses.io ● Initiative of OSM-US Chapter (Ian Dees &c.) ● Repository on Github of metadata for address open data – Potentially thousands of available data sets at local level in US & Canada ● Most information for US & Canda – Also some France (BANO), Netherlands & South Africa
- 20. SummarySummary OSM UK AddressesOSM UK Addresses ● Biggest geolocated open dataset for UK ● Crowd-sourced – if 2's a crowd ● Mainly ground truth surveys ● Patchy & Partial ● Not yet at critical mass ● Viral Share-alike licence (OdbL) – OSM as an EnablerOSM as an Enabler ● Large test dataset ● Knowledge & skills ● Wide range of tools covering address data management ● Community committed to Open Address data w/w
- 21. Acknowledgements ● Christian Quest, OSM-FR (BANO) ● Ian Dees, OSM-US (addresses.io) ● Simon Poole, OSM-CH (Address QA) ● Geofabrik, Karlsruhe (OSM Inspector) ● Will Phillips, Nottingham (OSM-Nottingham) ● Matt Williams (OpenPostcodeFinder) ● Harry Wood, London (discussions)
- 22. Acknowledgements ● Christian Quest, OSM-FR (BANO) ● Ian Dees, OSM-US (addresses.io) ● Simon Poole, OSM-CH (Address QA) ● Geofabrik, Karlsruhe (OSM Inspector) ● Will Phillips, Nottingham (OSM-Nottingham) ● Matt Williams (OpenPostcodeFinder) ● Harry Wood, London (discussions)