Maps! Maps! Maps!

Making it easier to discover (historical) maps in Wikimedia Commons

(And on other websites)

What do we need?

Store georectification metadata in Commons/Wikidata:

  1. mapping between pixels and latitude/longitude coordinates;
  2. pixel mask;
  3. Then, we can compute geospatial polygon of area depicted on map!

Many maps on Commons have been georectified with Wikimaps Warper

(or on other websites, such as David Rumsey or NYPL)

What did we do? (1/2)

We worked with 5 large repositories of maps:

  • Wikimedia Commons: lots of maps, some of them georectified with Wikimaps Warper;
  • Library of Congress: 5 million maps, more than 100,000 in Commons;
  • British Library: 35,000 georectified maps in Commons;
  • David Rumsey/Stanford: more than 40,000 georeferenced maps, thousands of them in Commons;
  • New York Public Library: 100,000s maps, many of them georectified, and many of them in Commons.

What did we do? (2/2)

  • Data specification (for Commons and outside of Commons)
  • Prototype microservice to georectify maps using this data specification;
  • Property proposals for WikiData;
  • Started identifying overlapping maps between map repositories.

What's next:

  • Get property proposals approved;
  • create a WMF Labs microservice that georectifies maps, based on those properties;
  • create WikiData items for all LOC maps in Commons (based on LOC metadata);
  • Identify overlapping maps, and get georectification data into Commons/WikiData.

Thanks!

This was a project by Bert Spaan, Matt Miller and James Heald, with help from Albin Larsson.

See GitHub for details and links: github.com/bertspaan/wikimania-hackathon-2019.