Re: AW: Recommendations: URIs

On 5/3/11 8:00 PM, Karen Coyle wrote:
> Quoting Antoine Isaac <aisaac@few.vu.nl>:
>
>
>> On "not exactly in those words", I'm not sure the point I wanted to make is entirely captured by the recommendations I can currently find, "Create URIs for library resources in good time" [1] or "Develop policies for RDF vocabulary namespaces" [2].
>
> I agree. We're thinking of doing some reorganization of the recommendations page, and I will work on the URI section to try to get more of this discussion into it.
>
> One thing that has me confused, however, when folks talk about assigning identifiers for library resources... if they mean "instance data" (which I interpret to be something like "bibliographic records" but in LD), that can't really be created until there are URIs for element sets and value vocabularies, true? So I'm not sure how we can focus on instance data first. Would instance data mean creating URIs for books and journals and WEM fragments?


Well, first I think previous comments included value vocabularies in the "instance data" here. If a library has a specific authority file, they should feel responsible for it.
As regards the metadata element sets, indeed the nascent recommendation is that the vast majority of libraries would not need to mint and maintain themselves URIs for classes and properties. They could (and even, should) re-use the classes and properties from already available ontologies. Ontologies which would be either produced by non-library actors, or produced by the much smaller number of libraries that would take care of developing library-specific LD metadata element sets (the latter being the point I'm trying to make in [1]

MARC is the analogy in the "classical" world, which comes to my mind (now tired, sorry). The huge majority of libraries should not take care of developing MARC, should they?

Antoine

[1] http://lists.w3.org/Archives/Public/public-lld/2011May/0015.html

Received on Tuesday, 3 May 2011 21:26:56 UTC