W3C home > Mailing lists > Public > w3c-rdfcore-wg@w3.org > June 2002

Re: procedure for determining reserved vocab

From: Patrick Stickler <patrick.stickler@nokia.com>
Date: Mon, 24 Jun 2002 09:59:00 +0300
To: "ext R.V.Guha" <guha@guha.com>, RDF Core <w3c-rdfcore-wg@w3.org>
Message-ID: <B93C9DE4.173C1%patrick.stickler@nokia.com>

On 2002-06-23 20:55, "ext R.V.Guha" <guha@guha.com> wrote:

> 
> Hi All,
> 
>   If memory serves me right, I was assigned the homework of writing
> the procedure for determining whether a uri corresponding to a node is
> from a reserved vocabulary.
> 
> The procedure needs to be able to determine whether the uri is reserved
> with respect to a layer. So, a vocab. item may be reserved wrt RDF but
> not OWL. It was also felt that it would be good if this procedure would
> work not just with W3C recos, but also languages from other such bodies.
> 
>  The procedure takes 2 arguments: (a) the uri (b) a layer name (or uri)
> 
>   We need to identify two things in the first argument.
> (i) whether it is in the reserved vocab.
> (ii) which reserved vocab. it is part of.
> 
> This can be done in one of three ways:
> (1) We invent a new uri prefix (such as log: so that a reserved vocab.
> item such as implies will have a uri such as
> log:www.w3.org/cwm#implies). The second part of the uri is the uri for
> the layer defining the term.
> (2) We use normal http uris and embed this somewhere in the middle of
> the uri. eg., http://www.w3.org/reserved/cwm#implies or
> http://www.w3.org/reserved/http://oasis.org/cwm#implies. This violates
> the principle that uris should be opaque.

Well, both (1) and (2) violate this principle, but at least the
violation of (1) is far less severe, and far more generic, so to
that extent (1) is to be strongly preferred to (2).

> (3) We use an appropriate encoding convention for the reserved vocab.
> item's Qname. e.g., instead of "implies", we use Log_CWM_implies. While
> this violates its own set of principles, it does have the advantage of
> not requiring a new namespace.

This cannot work since qnames are not present in the graph, and I'm
presuming that the 'darkness' must be determinable based on the graph.

So unless we are also employing some mechanism of the graph syntax
to flag unasserted triples, any information embodied in qnames
(namespace prefixes) will be lost.

Though, if you mean that the encoding is part of the local name, then
I don't see then how (3) differs any from (2).

So (1) seems the only reasonable option.

A possible starting point for the definition of the log: URI scheme
could be

http://ietf.org/internet-drafts/draft-pstickler-uri-01.txt

Simply replace 'uri:' with 'log:'.

Thus, just as URI reification is an encapsulation of a URI, so
also would darkening simply be an encapsulation of a URI, and
that would permit the use of the same term in both asserted
and unasserted statements.

Question: do we want to call it 'log:'? Does that really capture
the notion of unassertedness or darkness?

> Once we decide which of these three roads to take, the procedure is
> quite trivial.

Well, the decision procedure for the WG might be trivial, but the
application of that decision could be anything but trivial (though
I don't think it should be).

Cheers,

Patrick


> guha
> 
> 
> 
> 
> 
> 

--
               
Patrick Stickler              Phone: +358 50 483 9453
Senior Research Scientist     Fax:   +358 7180 35409
Nokia Research Center         Email: patrick.stickler@nokia.com
Received on Monday, 24 June 2002 02:54:50 EDT

This archive was generated by hypermail pre-2.1.9 : Wednesday, 3 September 2003 09:49:21 EDT