- From: Jonathan Borden <jborden@mediaone.net>
- Date: Tue, 20 Mar 2001 07:22:22 -0500
- To: "Dan Brickley" <danbri@w3.org>, "Jan Grant" <Jan.Grant@bristol.ac.uk>
- Cc: "rdf interest" <www-rdf-interest@w3.org>
Dan Brickley wrote: > > <wn:Person xmlns:wn="http://xmlns.com/wordnet/1.6/Person" > xmlns:foaf="http://xmlns.com/foaf/0.1/" > foaf:prisonerCode="6" /> > > Yep, though there's no huge social difference between assigning a > number/ID as someone's "URI" versus relating them to a > number/mailbox/homepage/etc. Actually I think there is a subtle but important difference. When you really need to keep track of someone (for good reason), for example healthcare, you assign an identifier code ... but it is _critical_ that the distinction between the person and the person you _assume_ is identified by the code be maintained. For example a surgical booking system. Someone may schedule an operation based upon name and identifier (common practice) yet it is entirely possible to have 2 people with the same name in adjacent rooms and hence type the wrong _identifier_ into the system. You really can't blindly go ahead and perform surgery on an identifier: that's how those disasters where the wrong kidney etc gets removed _always_ a misidentification. So the point is that you treat the person as an anonymous node and use an algorithm to equate (e.g. daml:equivalentTo) the anonymous node with a node identified by the identifier. This is a real world and very real wrinkle on the error of naming an anonymous node with a generated identifer, rather than addressing the anonymous node (e.g. with an XPath or RDF query string). > > A lot of what we want URIs to do for us, we can do with other forms of > identifying expression (typically built from URIs... :) so there's no > escape, even for things that don't have URIs... > Just remember than not all URIs are names, some are addresses. Perhaps this is another reason to standardise a query syntax: it would provide an agreed upon method for _addressing_ as opposed to _naming_ anonymous nodes. For example such a syntax might serve as the fragment identifier in an RDF media-type. Jonathan Borden The Open Healthcare Group http://www.openhealth.org
Received on Tuesday, 20 March 2001 07:21:09 UTC