Greetings, all.
Many thanks, Monica and Andrew, for the pointers to the JISC identifiers strand (I've somehow failed to register that strand properly, though I'm aware it's been quietly getting on with things for a while), and for the pointers to the ANDS guides, which I'll work through.
On 2010 Oct 1, at 16:31, John A. Kunze wrote:
> --- On Thu, 30 Sep 2010, Norman Gray wrote:
>>> EZID (http://n2t.net/ezid)
[...]
>> In particular, is it clear whether this is fundamentally different from
>> purl.org? Obviously, it's a different implementation of the same basic
>> idea, and it uses ARKs, but is there an argument why one should use this
>> rather than purl.org, in terms of the long-term persistence model, or
>> curation style, or...?
>
> EZID is identifier-scheme-agnostic. Its designers believe resolution
> and administration should be decoupled from identifier syntax. The tools
> available for ARKs, DOIs, Handles, PURLs, URNs, and even URLs -- some
> well-known, others less well-known -- all essentially provide a way to
> associate metadata with identifiers.
(i) Ah yes, schemes. Myself, I'm broadly persuaded by the use-URLs-for-everything dogma: while scheme-agnosticism is a Good Thing, URLs are very conveniently usable now, and will (with probability near one) have a well-defined mapping to whatever replaces HTTP and the web (as you note in draft-kunze-ark-15). But that's not the issue right now.
In any case, for the application I'm currently interested in, I'm specifically interested in long-term URIs, rather than identifiers in general.
(ii) Thanks, John, for emphasising the centrality of identifier metadata for this. I tend to use purl.org as a touchstone for reading about things in this area, and for all its virtues, identifier metadata isn't as front-and-centre for purls as it is for other mechanisms.
> Another key shared concept is a metadata element known roughly as the
> resolution target(s), which serves to populate an ancillary "resolver"
> (database) that specializes in forwarding incoming requests.
If I'm understanding you correctly, that's a somewhat formal way of saying that all identifiers schemes have a thing that the identifier redirects to, whether that be a dataset, a PDF, or the concept of 'justice'. Formal is good, by the way; I'm just checking I'm getting the right end of this particular stick. Can you point me to somewhere that formalism is worked out ('target' isn't present in draft-kunze-ark-15)?
> One of the motivations for UC3/CDL in creating EZID is that we will need
> to deal with objects and identifiers from many different sources. EZID
> obviates the need to require that everything conform to one scheme.
A true-blood url-for-everything dogmatist would say 'mint a URL for it!', but I do take the point.
Best wishes,
Norman
--
Norman Gray : http://nxg.me.uk
|