How to implement human readable PIDs?

By human readable PIDs, we mean that the information accessed via the identifier is readable, rather than the identifier itself. The main requirement for human readable PIDs is that each identifier takes users to a webpage that describes - and allows access to if possible - the item(s) to which the PID refers. This can be referred to as a "landing page". It is recommended that the identifier itself is "opaque", i.e. does not contain human-readable words that contain specific information, Because these PIDs need to be managed for the long term and should not be changed or deleted, including meaningful information that can change over time should be avoided.

They can be implemented using internally created URIs or identifiers generated from an external provider. The identifiers should be displayed on the page as links. Any PID including Handles, URIs, DOIs and ARKs can be implemented as human readable PIDs.

The British Library, Natural History Museum and Royal Botanic Garden Edinburgh case studies all describe different methods of implementing human readable PIDs. The British Library's repository implements human readable DOIs which are displayed with landing pages with access to the files.

The Natural History Museum has a similar use of DOIs for its research datasets on the Data Portal. The records for its specimens are available in a human readable format via the Data Portal, which is a specified "data" domain within the Museum's website. They are also available via an API for machine access, but one type of access does not require the other. In practice many organisations provide a human readable access and machine-readable access is developed later, but it can often be easier to develop both simultaneously as the Natural History Museum and National Gallery did.


    Implementing human readable PIDs does not require much effort once you have web pages assigned to each item displaying either the item or metadata about it. The scale of the effort in creating those pages or how much of that work has already been done may inform decisions on the level at which you are able to assign identifiers. For example, whether you want to consider having a PID for each image will be a different conversation if you need to create metadata and webpages for three million images than for three thousand images.

    There may be a cost associated with generating the identifiers if you are to use a charged third party service such as DOIs. Again, depending on the user options available to you, cost may determine how many you are able to create for schemes that charge on a scale relating to the number of PIDs assigned.