Workbook on Digital Private Papers > Administrative and preservation metadata > Persistent identifiers
Persistent identifiers
Adding persistent identifiers: when to identify?
Labels to identify an object can be added at various times during an object's life, and an object could have several identifiers, simultaneously or consecutively, for different purposes, just as individuals are allocated identifiers by the various systems in which they participate.
Key stages at which an identifier might be added include:
Original object creation
There may come a time when donors provide materials that already possess persistent identifiers. Whether or not these identifiers are retained may depend on the scheme selected by the donor; there might be an ongoing cost associated with doing so. It is possible to associate more than one identifier with an object, so one option may be to use the donor's identifiers in addition to identifiers allocated by the repository. Quite often in paper archives indexing references employed during the records' active life are retained alongside new cataloguing arrangements. If this is done, it should always be clear to the user that the earlier identifier has been superseded by the current, repository-assigned one.
Object creation via migration
If adopting a migration preservation strategy, then the repository may also need to create one or more representations of the original object in more accessible formats. These objects will need their own identifiers and their relationship with the intellectual entity they represent, and thereby with other representations (especially the representation from which they derive), must be expressed.
Object ingest
While archives arrive without persistent identifiers, it seems that ingest (where the repository takes intellectual control of the archive and processes it for preservation) is the most appropriate time to allocate persistent identifiers to the objects to be preserved and for the metadata created about them. If the objects are to be kept in a local dark archive, the identifiers need only be unique within that environment. In a traditional archival context, identifiers (such as reference codes or shelfmarks) are not usually applied at item level until an archive is arranged or described, although a unique accession number is usually assigned to collections or instalments of a collection when they are taken in. In the case of digital archives, some form of item-level identifier must be allocated sooner than this so that preservation metadata can be associated with objects and the original order of an accession can be recorded.
Object dissemination
When Dissemination Information Packages (DIPs) for reader access are prepared, it may be that objects are moved from the restricted dark preservation archive into a more open repository. If publishing objects to an online repository, or allowing others to harvest objects (or their metadata) for use in their own repositories, the repository must ensure that the identifiers it uses are globally unique. It is this global uniqueness that many of the PID schemes for digital objects aim to provide. It may be useful to employ identifiers which conform to one of these PID schemes even when objects are held in a local dark archive, in order to ease the transition from the closed to the open environment.
Events and agents
Events which need to be recorded must have identifiers as part of their metadata, as must agents associated with events. This allows the repository to associate events with relevant agents, and events with objects to which they pertain.
