1 of 10

Experiences of a user of a range of identifier services

​

Simon J D Cox

2 of 10

Linked Data and the Research Web relies on Persistent Identifiers

3 of 10

A bit of history

  • Cool URIs don't change �TimBL 1998

​

  • W3C policy:
    • ‘no need for anything other than HTTP URIs
    • Conflates identifier and locator
    • PID = URI + strong governance

​

  • In practice … much link-rot (404) even from respectable organizations
    • BBC, UKEA, NEII, CSIRO, GA, ABS

​

4 of 10

Handle

  • CNRI 1995
  • Compatible with, but not dependent on DNS

5 of 10

URN

Identifier, not locator

​

Local resolution patterns

6 of 10

PURL

  • OCLC 1998
  • Friendly API/UI to manage redirects

​

  • OBO has a PURL service for ontologies
  • OCLC’s generic PURL service http://purl.org never transitioned from research 🡪 production
    • Offline for 18 months in 201x’s
    • Archived at https://purl.archive.org/
    • Led to creation of w3id.org consortium

7 of 10

ARK

  • Interesting, some uptake

8 of 10

DOI, ORCID

  • The most successful persistent URI system – why?
    • Solid business model
  • Backed by publishers
  • Built into academic credit& rewards process

9 of 10

Datacite, IGSN, ROR

  • Needed for research web
  • Weak business model, so have struggled to gain momentum and maintain a viable process

​

10 of 10

Issues

  • Business model is key to long term viability
  • PIDs for non-digital resources

​

​