W3C home > Mailing lists > Public > semantic-web@w3.org > April 2008

Person Identifier

From: Bent Rasmussen <incredibleshrinkingsphere@gmail.com>
Date: Mon, 21 Apr 2008 00:00:21 +0200
Message-ID: <2D8F14A537614464A03D814A159FEE9F@BentPC>
To: <semantic-web@w3.org>
Some questions I've been pondering and which you may have some answers for...

Is there a best practice for naming people? I suppose it isn't particular to people, but it appears to me that to use an email address as an identifier is rather arbitrary as an identifier. If everything important should have a URI, shouldn't this URI have some consistency? An email address may be taken over by someone else. So may a domain name. In the end even a name could change, but the thing that was denoted (and changed over time) should have a consistent identifier, I'd suppose. In the face of changing addresses this appears to break down, to me.

Of course one can also apply a weaker "extensional" meaning of identity and compare for equality on a property level.

And how about changing properties - do we just create new definitions once properties change? How about the old definitions and ambiguity (if not incompatibility)? Do we use a special URI scheme for assigning temporal definition?

Also, is the practice of structured identifiers not creating somewhat of a microverse outside the grasp of RDF? For example

http://www.rfc-editor.org/rfc/rfc4151.txt

It appears to me that the semantics are creeping into identifiers. Then again they can just be seen as parseable conventions for naming things and semantics can be extracted from them.

I tend to favor protocol-independent URIs (URNs) because they really are just identifiers and do not "age" with protocols, so to speak - or worse, become misleading over time.

(Inspired by "Cool URIs dont' change")

- Bent
Received on Sunday, 20 April 2008 22:01:00 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:42:04 UTC