OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.


Help: OASIS Mailing Lists Help | MarkMail Help



   RE: [xml-dev] Assumptions about URIs

[ Lists Home | Date Index | Thread Index ]

>Bent Rasmussen wrote,
> > Perhaps the central question would be:
> >
> > - "Does the URI 'http://www.w3.org' identify an
> >  organisation called W3C or a location on the Web?"
>I think I must fall in camp D, because although I'm temperamentally
>inclined towards camp B, my answer to this question would be
>"Potentially both, and a great many other things besides".

Interesting. But does it make sense not to choose? As an example let's say

(a) <urn:isbn:...> <urn:concept:author> <http://example.com/mister.x>

"the author of a book (...) is the location 'http://example.com/mister.x' 
but is also the person which I thought of as I picked the URL."

I mean, wouldn't you have to choose between

(i) "it's a person, not a location, damnit", or
(ii) "it's a location not a person, damnit"

:) Otherwise you'd be making two statements, wouldn't you? If you choose (i) 
then you disregard the semantics of the namespace, i.e. for example that the 
TEL scheme refers to telephone numbers or that the HTTP scheme refers to 
hypertexts at locations. This seems important, for example some vocabularies 
are used to make statements about home pages while other vocabularies are 
used to make statements about things that users "think of" as the resource 
of a URI. So some seem to say that HTTP/TEL/..., it doesn't matter because 
we can say the resource of "tel:2342343" is a person or the resource of 
"mailto:x@x.x"; is a space craft; others say that HTTP is for one thing, TEL 
is for another etc. Don't addresses in the TEL name space *necessarily* 
identify phone numbers?

So, some systems have statements about keys that are themselves data and in 
addition that data exists in a context, a name space. The context implies 
types (telephone numbers, books, hypertext locations, etc.) and the keys 
imply instances (certain telephone numbers, certain books and certain 
hypertext locations).

In other words we've got two sets of data, the data that is apparent to a 
knowledge representation system (statements about the keys) and the data 
that is not apparent to the knowledge representation system (the keys 
themselves and their contexts).

One could probably use HTTP to specify many other types; e.g.

- "http://www.example.com/tel/2345345";

But this goes against practice and recommendations it would seem. And since 
URNs may also have network retrievable representations it might be better to 
use that. Either that or the name space context is truly irrelevant to the 
semantics of the resources. And its still interesting that the key is itself 
data, but that this data is inaccessible to the system unless some special 
treatment is going on.

Of course one can use things like rdf:type to assert the type of the 
resource and thereby override or contradict the authority and the creator -- 
is this the solution?

>Miles Sabin                                     InterX
>Internet Systems Architect                      27 Great West Road
>+44 (0)20 8817 4030                             Middx, TW8 9AS, UK
>msabin@interx.com                               http://www.interx.com/

MSN Photos is the easiest way to share and print your photos: 


News | XML in Industry | Calendar | XML Registry
Marketplace | Resources | MyXML.org | Sponsors | Privacy Statement

Copyright 2001 XML.org. This site is hosted by OASIS