Registries were a hot topic at last week’s Dublin Core 2005 Conference in Madrid, Spain. One highlight was the gathering of the registry-heads: the DCMI Registries Working Group meeting on Wednesday.
Wednesday afternoon (and early evening–this is Spain, after all) at DC2005 was the DCMI Registries WG, where all the registry-heads gathered–it was a very interesting session indeed, despite the creeping sleep deprivation overcoming us all by Wednesday. Harry Wagner and Rachel Heery have been heading the group; Harry announced that he’d be moving on to other projects at OCLC and would be giving up his role on the WG.
- Simplify discovery of terms and specify the relationships between terms
- Promote the reuse of metadata terms
- Provide a trusted source of information about terms (and persistent URIs)
After some discussion, several attendees gave brief presentations on current registry implementations, most using the DCMI Registry software.Karen Rollitt discussed a current New Zealand implementation of the software (see http://dcregistry.natlib.govt.nz/dcregistry/). A highlight of this implementation is the Maori version of Dublin Core. Karen reported that the National Library of NZ plans to translate more terms into Maori, and add versions using other Pacific Island languages as well as add usage examples in those languages.
Harry mentioned that the DC copyright should appear on implementations. This got a bit of discussion going since the software is supposed to be open source and the content not the same for other implementations–language versions, for instance, are certainly not owned or copyrighted by DCMI. The conclusion was that we should have someone with more knowledge of licenses look at this (Stuart?), and suggest updates to the statements and requirements for those using the software. One complication that came up was the question of how would this work in the context of an application profile, where you might want to refer to a DCMI term.
Mitsuharu Nagamori, University of Tsukuba, discussed the Tsukuba implementation (see http://fiord.ulis.ac.jp/dcregistry/index.html). This is a significantly different implementation, registering ca. 50 non-DCMI schemas. One question I had that I didn’t have a chance to ask (too busy taking notes!) was whether permission had been sought to register schemas that were not owned or maintained by Tsukuba. The GEM schema is registered, for instance, and I can’t imagine that permission was sought or given for that. This is the same problem that the Usage Board came up with when we were considering registering tokens to describe vocabularies, and it seems unlikely that it was considered by the Tsukuba group.
Mitsuharu did mention that they have developed a software tool generator, based on metadata schema. It’s not clear to me exactly what that tool does, but it might be worth looking into. Mitsuharu also noted two lessons learned:
- Need to add relationship between term and schema, and term and term.
- Need to improve user interfaces, which they felt were difficult to scale.
Thomas Fischer of Goettingen, briefly discussed his experiences with a Registry Lite implementation. They were mostly interested in language functionality and thought the software installed well.
[NOTE: As I was looking at this implementation, it struck me that usage of the term “classification type” for elements, element refinements, etc., is not going to work for vocabularies.]
One big issue that arose in the discussion of these implementations is versioning, still potent despite Tom’s work to rationalize the DCMI practices. Versioning is certainly an issue for schema translations and looms large for vocabularies as well.
Rachel Heery reported briefly on 8th Open Forum on Metadata Registries, Berlin, 11-14 April 2005. There was apparently much concern that the DCMI registry efforts were largely ignoring the family of ISO 11179 Standards, though all agreed that we should be collaboration. Gail Hodge agreed to try and contact the folks responsible for those standards and see whether there were opportunities. Apparently efforts in the past to do so have not been successful but hope springs eternal.
Rachel also gave an update on the JISC Metadata Schema Registry (see: http://www.ukoln.ac.uk/projects/iemsr/). This implementation uses a separate piece of software developed at Bristol. She also pointed to some usage scenarios they developed (see: http://www.ukoln.ac.uk/projects/iemsr/wp2/function/#sec3). JISC also has some demonstrators and funding for another year. They will be looking at business cases and also want to improve machine to machine interfaces to the registry and navigation as registry content scales.
The group came up with some areas of potential collaboration:
- User requirements
- Business case
- AP data model
- shared test data
- peer review
- joint dissemination
There was also some enthusiasm for exchanging information on progress and issues, via this blog and perhaps a shared wiki.
No comment yet