Re: National Identification Number URIs ( NIN URIs )

Kingsley and all, hello.

On 2010 Mar 8, at 14:44, Kingsley Idehen wrote:

> The original question also made a reference to DBpedia namespace, which is what triggered the Linked Data discussion etc..

Indeed, and a useful discussion it is, too.  I was just thinking about the IDs, and the implied question about whether such a standard for IDs already existed.

> In this scenario the canonical records are best served with URNs rather than HTTP URIs at the core DBMS level. They can we published to HTTP networks in line with HTTP based Linked Data best practices (which doesn't define Linked Data the concept general, just how you apply the concept to HTTP Networks such as the World Wide Web).

At one level this is an almost self-evident remark: if you have a compact ID for you an entity, then it almost certainly makes sense to store that ID in your database, rather than the full HTTP URI that you'd publish outside your system boundary.

A key thing about (what I believe to be) your position here is that there's a clear distinction between inside and outside the system boundary.  However there can be other boundaries, further out.

David Booth's excellent [1] "Converting New URI Schemes or URN Sub-Schemes to HTTP" is a discussion of the mapping between URNs or other non-HTTP identifiers, and (cool) HTTP ones.  I summarise it as "it's OK to have both", which might take some of the heat out of perennial HTTP-or-not discussions.  It implicitly draws a porous boundary between those who can grok the non-HTTP URIs (and get enough payoff to take the trouble), and the fully external world-wide bit of the web.  Neither need exclude the other.

Or: the procedure that manufactures an ID string is distinct from the punctuation that lets it be manipulated in the world.  Which isn't a terribly profound remark.

All the best,

Norman


[1] http://www.dbooth.org/2006/urn2http/

-- 
Norman Gray  :  http://nxg.me.uk

Received on Monday, 8 March 2010 15:20:26 UTC