In 2017, the Metropolitan Museum of Art made history by releasing 375,000 public domain images under Creative Commons Zero. It was the largest open access release by any museum at the time.
But images are only half the story. Without structured metadata — machine-readable information about what each artwork depicts, when it was made, who created it — those images are hard to find, harder to connect, and impossible to query at scale.
That’s where Wikidata came in.
Building the Bridge
Starting in 2019, I worked with the Met’s collections team to build a pipeline connecting their internal catalogue to Wikidata. The goal: create Wikidata items for every artwork in the collection, complete with structured statements about creator, date, material, dimensions, and subject matter.
By mid-2020, we had created over 14,000 Wikidata items for Met artworks. The Met’s General Manager of Collection Information, Jennie Choi, built on this foundation — adding depicts statements to 4,000+ items, uploading 5,000+ new images to Commons, and mapping the Met’s entire subject keyword vocabulary to Wikidata.
What Changed
The impact was immediate:
- Artworks previously invisible to search engines became discoverable
- Wikidata queries could now answer questions like “show me all 17th century oil paintings depicting cats” across the Met’s entire collection
- Voice assistants started returning Met artwork data from Wikidata
- Google Knowledge Graph panels for artworks sourced their structured data from Wikidata
But the biggest change was cultural. When Jennie Choi — a senior museum professional — wrote about her Wikidata experience for the Wiki Education Foundation’s blog, she sent a powerful signal to the museum world: Wikidata skills are professional development for cultural heritage workers.
The Tools That Made It Possible
The work relied on several key tools:
- QuickStatements: bulk Wikidata editing via simple text commands
- OpenRefine: data cleaning and reconciliation with Wikidata
- Pattypan: bulk image upload to Wikimedia Commons
- Custom crosswalk databases: mapping museum vocabularies to Wikidata items
None of these are museum-specific. They’re general-purpose tools adopted by the Wikidata community. The innovation was applying them to cultural heritage at scale.
Lessons for Other Institutions
- Start with a pilot. Pick one collection area, create 100 items, learn what works.
- Map your vocabulary to Wikidata. Before creating items, know how your categories map to Wikidata concepts.
- Hire or partner with a Wikimedian. Having someone who understands both museum workflows and Wikimedia community norms is essential.
- Plan for sustainability. Museum catalogues change. Build workflows for updating Wikidata when your records change.
- Celebrate the wins. When you see your artwork illustrating Wikipedia in 30 languages, share that story internally. It builds momentum.
The Met’s journey shows what’s possible when cultural institutions embrace linked open data. Not as a one-time project, but as an ongoing partnership with the Wikimedia community.