Dublin Core Challenges

I am writing a paper on describing services with Dublin Core. As I do this, it seems smart to keep a list of issues that I and/or others have had with regards to Dublin Core. Such a list might include things I would like to see changed in Dublin Core.

1. Rights does not have a Jurisdiction refinement. Something may be in Copyright, but copyright terms vary from jurisdiction to jurisdiction.
2. The Identifier element does not have a common set of refinements.
3. Names do not have a way to define an order — though this can be partly to blame on software implementations.
4. Names do not have a way to define differences in family vs. given names and similar concepts in other cultures.
5. Collection type vocabulary is not part of the official standard. (Why not make it part of the official standard? Its optional just like the DCMIType vocabulary, but would get more visibility and traction.)
6. Accrual vocabulary is not part of the official standard. (Why not make it part of the official standard? Its optional just like the DCMIType vocabulary, but would get more visibility and traction.)
7. Other library of congress subject or type vocabularies are not available (or maybe they shouldn’t be)
8. There is no current way of saying BCP47 is the standard for language tag use.
9. Services do not have a way of overtly identifying the language of the user interface in addition to the language of the data acted upon (input) and language of the output.

Lissonet has some interesting discussion in her dissertation on the issue of limitations in Dublin Core.

These include:

• great examples for dates (epochs).
• great examples where place names have not become part of a formal registry.
• issues with format, medium and suggesting that those should be part of description.
• great examples about Rights permissions.

While I don’t agree with all of the exampels Lissonnet puts forward, I do think that some of these are worth exploring in other contexts outside of the where she is primarily working. Some of these issues are re-stated in the short paper. Simons I believe mentions the issue with the names. While person id’s and identifier types are acknowleged in various places such as schema.org and Dublin Core forums and community groups.

References