Abstract
The identification of information objects has always been important with library collections with indexes having been created in the most ancient times. Since the digital age, many specialised and generic persistent identifier (PID) systems have been used to identify digital objects. Just as many ancient indexes have died over time, so too PID systems have had a lifecycle from inception to active phase to paralysis, and eventually a fall into oblivion. Where the indexes within the Great Library at Alexandria finally succumbed to fire, technology change has been the destroyer of more recent digital indexes.
We distil four PID system design principles from observations over the years that we think should be implemented by PID system architects to ensure that their systems survive change. The principles: describe how to ensure identifiers’ system and organisation independence; codify the delivery of essential PID system functions; mandate a separation of PID functions from data delivery mechanisms; and require generation of policies detailing how change is handled.
In addition to suggesting specific items for each principle, we propose that a platform-independent model (PIM) be established for persistent identifiers – of any sort and with any resolver technology – in order to enable transition between present and future systems and the preservation of the identifiers’ functioning. We detail our PID system—the PID Service—that implements the proposed principles and a data model to some extent and we describe an implementation case study of an organisation’s implementation of PID systems that implement the Pillars further but still not completely.
Penultimately, we describe in a Future Work section, an opportunity for the use of both the Pillars and the PIM; that of the World Wide Web Consortium’s Permanent Identifier Community Group who is seeking to “set up and maintain a secure permanent, URL re-direction service for the web”.