RE: Review of draft finding on URNs, Namespaces and Registries

[Context: "URIs are names", and "URIs should not have a hierarchical
part"]Fernando Franco writes:

> Why should names be hierarchical, at all?Noah Mendelsohn replies:

> I think there are many good reasons for this, and there's a long history
> of doing it.  Names are often arranged hieararchically to make it easier
> to deal with collections, to infer from the name something about the thing
> that is named, etcLet's imagine somebody in a kitchen:

      "Hey, you, /animal/cat/felix , get outta that fish!"

Noah...it wouldn't work, right? :)

In fact, if we used hierarchical URI's, we would be getting exactly the same
problem that we got with XML. We would be able to do just one kind of
classification ( /animal/cat/felix ), and not others (
/consumer/furry/felix ). So yes, the problem goes back all the way to the
same reason why we don't use XML alone, and by which RDF exists at all:
tree, not graph..

What you mention, *was* sensible (in a web 1.0 context). I have made use of
it, just as you did.
But in a semweb context, we have now a full metadata level. If you want to
know properties (structure, whatever) of resources (including collections),
you can get them straight away. You no longer need to attempt infering
anything at all from it's name.

If there are other reasons why URI's should be hierarchical, however, I'd
like to hear them, least of all would we break something up.

Respectfully,
Fernando Franco

Received on Tuesday, 22 August 2006 07:12:55 UTC