How to implement machine readable PIDs?

By machine readable PIDs, we mean PIDs where the metadata about the resource to which the PID refers can be processed by a machine. To implement machine readable PIDs, items within the collection need to have information associated with them which can be processed or rendered in automated ways. This can be done by making metadata available in a machine readable format, e.g. JSON or RDF.

Any PID including Handles, URIs, DOIs and ARKs can be implemented as machine readable PIDs so long as the metadata is in a machine readable format.

The National Gallery's PID implementation includes both a human readable and machine readable aspect. When a machine requests a PID, the landing page metadata is returned with ‘.json’ at the end of the URL rather than .html or another webpage language. JSON is a machine readable format, if a human users wants to view a JSON version they can change the URL manually to see the JSON information. The data is taken from the collection management system and transformed by the CIIM middleware solution to a variety of formats.

The NHM expresses the metadata on its data portal in several formats including machine readable RDF and it is accessible via an API. Both of these applications were driven by integrating with linked open data collections and in the NHM's case integrating with the CETAF Stable Identifiers initiative. Both are described in more detail in their respective case studies (available here and here).


    Machine readable PIDs do not have to cost anything additional to human readable PIDs to implement. However all technology services require some maintenance, and therefore a SPARQL endpoint or an API may need to be updated and maintained periodically. It does not cost more to implement DOIs with content negotiation, but the content negotiation needs to be developed and maintained internally.