Structured data for GLAM-Wiki/Reconcile

Structured data for GLAM-Wiki

• Home
• Introduction to structured data
• Preparation
  • Connect with the Wikimedia community
  • Check what is already on Wikimedia projects
  • Clarify copyright status
  • Prepare data and media
  • Reconcile
• Upload
• Activity around materials
• Measuring impact
• Data roundtripping

People, places and other concepts in your data need to be 'matched' with their corresponding Wikidata items.

Reconciliation: matching people, places, keywords and other concepts with Wikidata

When you upload structured data to Wikimedia projects from an external dataset, at some point it will be necessary to 'translate' or reconcile your data items to their equivalents on Wikidata. Most frequently, reconciliation happens for the following types of data:

• People and organizations (or 'agents') – often creators of works (painters, sculptors, architects, photographers, authors, editors, publishers...)
• Place names – for instance names of municipalities or countries
• Keywords – terms and subjects with which collections are described

Example

Cornelis Martinus Vermeulen: Van der Borght, Nicolaes (17e eeuw); koopman te Antwerpen, 1703. Collection FelixArchief

The FelixArchief (municipal archive of Antwerp, Belgium) has contributed a set of media files to Wikimedia Commons in 2019. Their source data set, exported from their collections database, contains names of people that are formatted in the following way:

Vermeulen, Cornelis Martinus
Michiels, Jean Baptiste
Lauwers, Frans
Ghémar, Louis
Van Veen, Otto
Mostaert, Jan

Note that the institution uses a Lastname, Firstname notation for names of people.

The reconciliation process is the specific (and crucial) step where the corresponding Wikidata item or entity is looked up:

Original text string	Corresponding Wikidata item
Vermeulen, Cornelis Martinus	Cornelis Vermeulen (Q2997496)
Michiels, Jean Baptiste	Jean Baptiste Michiels (Q21396012)
Lauwers, Frans	Frans Lauwers (Q42899838)
Ghémar, Louis	Louis-Joseph Ghémar (Q16214059)
Van Veen, Otto	Otto van Veen (Q785355)
Mostaert, Jan	Jan Mostaert (Q577279)

Not all people, places, keywords in your data will already have Wikidata equivalents. It is OK to create new items. Be thorough during the reconciliation, though. Creating many duplicate Wikidata items must be avoided, as these cause a lot of cleanup work for the Wikidata community!

How to reconcile data with Wikidata

Several tools and software environments are able to reconcile (external) data with corresponding Wikidata items. These tools are suitable for different circumstances. When you are preparing collections data for an upload to Wikidata and/or Wikimedia Commons, OpenRefine is probably the most powerful tool to use.

• OpenRefine and its Wikidata reconciliation feature
• Mix'n'match, an online tool for crowdsourced reconciliation (Manual)

Developers - API