W3C home > Mailing lists > Public > semantic-web@w3.org > January 2007

RE: An RDF Reification Syntax Idea

From: Michael Schneider <m_schnei@gmx.de>
Date: Wed, 31 Jan 2007 22:14:20 +0100
Message-ID: <45C106AC.4060208@gmx.de>
To: adamsobieski@hotmail.com
CC: semantic-web@w3.org

Hi Adam!

Adam wrote:
> It occurs that in all the things that can be resources, one that is
> missing is a statement residing in another document.
> 
> I propose for consideration a syntax convention resembling:
> 
> "http://yourdomain.com/yourdocument.rdf!123" to indicate that the
> resource I wish to discuss or describe is the 123rd triple in the
> triples representation of the document at that address.

What you suggest here is to reference an RDF triple by its position 
within its containing RDF-graph's /serialization/. An (abstract) RDF 
graph is meant to be a /set/, and so its triples do not have any 
specific order. The order within the document does not have any 
RDF-related relevance: Every document consisting of a permutation of the 
triples will be deserialized into exactly the same RDF graph.

So, your proposed URI format will loose its usage in the moment when the 
document is deserialized into an RDF graph. That is different from an 
URI in '#'-notation: A #-URI is (by convention) used to reference some 
resource (or its description) within an RDF graph, independent from any 
concrete serialization syntax: After deserializing some, say, RDF/XML 
document, it is still perfectly clear, what resource is referenced by 
such a #-URI.

But even if you are really interested in referencing the triples within 
a concrete RDF document, you often cannot count on the order of its 
content. Say, you have the following document URL:

   http://www.example.org/foo.rdf

But, perhaps, this document does not really exist as a static file at 
server side. Instead, the (semantic) web server might generate it 
dynamically in the moment when it is accessed, by assembling its content 
from information within some database. You cannot expect the database 
owner to store additional ordering information within its database, 
because such an ordering is, as I pointed out above, of no relevance 
from an RDF point of view. And you also cannot expect the applied 
assembly algorithm to behave deterministically, so it is not warranted 
that it will always produce exactly the same document.

Best regards,
Michael
Received on Wednesday, 31 January 2007 21:21:11 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 1 March 2016 07:41:55 UTC