[oslc-core] Guidance on URI design for RDF/XML representations

[Arthur Ryman]

Ian,

+1

Although in RDF a property is simply a URI, in practice there are benefits 
to using URIs that can be represented as QNames since this also makes 
other formats (SPARQL, Turtle, JSON) more readable.

However, rather than cite just QName which is an XML concept, it would be 
better to also cite the guidance in the related RDF specs. For example, 
SPARQL 1.1 defines local names via the grammar rule PN_LOCAL [1]. Also, 
the Turtle grammar defines local names via the grammar rule PN_LOCAL [2] 
which looks identical to the SPARQL rule.

Therefore we should recommend that RDF terms should be built up from an 
HTTP URI prefix concatenated with a local name that satisfies the PN_LOCAL 
grammar rule, AND the local name should also satisfy the XML LocalPart 
grammar rule. [3]

I'd need to read the rules more carefully to say if in fact LocalPart is a 
strict subset of PN_LOCAL. The recommendation should be to use local names 
that work in all the important formats.

[1] http://www.w3.org/TR/sparql11-query/#rPN_LOCAL
[2] http://www.w3.org/TR/turtle/#grammar-production-PN_LOCAL
[3] http://www.w3.org/TR/REC-xml-names/#NT-LocalPart

Regards, 
___________________________________________________________________________ 

Arthur Ryman 

DE, Chief Architect, Reporting &
Portfolio and Strategy Management
IBM Software, Rational 

Toronto Lab | +1-905-413-3077 (office) | +1-416-939-5063 (mobile) 





From:   Ian Green1 <ian.green at uk.ibm.com>
To:     oslc-core at open-services.net, 
Cc:     Dominic Tulley <dominic.tulley at uk.ibm.com>
Date:   10/11/2013 06:22 AM
Subject:        [oslc-core] Guidance on URI design for RDF/XML 
representations
Sent by:        "Oslc-Core" <oslc-core-bounces at open-services.net>



What is the guidance on how RDF vocabulary terms should be chosen so as to 
ensure they are directly representable in RDF/XML (eg see [1])? 

For example, the triple 

        <http://example.com/req> <http://example.com/ns/1> "value of 
property 1". 

can't be directly represented in RDF/XML because one can't write the 
property "http://example.com/ns/1" as an XML QName.   (OSLC V2.0 
representations don't admit reification of RDF properties in general). 

I ask because we're allowing users to define their own vocabulary terms 
and we're struggling with how to express these constraints to the 
end-user, as well as concerned that other applications and vocabulary 
designers don't consider all RDF resource formats.  Since OSLC V2 requires 
RDF/XML format, it is unlikely today that OSLC consumers & providers will 
encounter RDF that can't be represented in RDF/XML, but this doesn't seem 
like a robust position as we move into V3. 

My first stab at the definition is "Each and every user-supplied RDF URI 
Reference MUST be representable as an XML QName".   (This is a sufficient 
condition but it might not be a necessary one.)   

best wishes,
   -ian 

[1] 
http://lists.w3.org/Archives/Public/www-rdf-comments/2001JulSep/0124.html 


ian.green at uk.ibm.com (Ian Green1/UK/IBM at IBMGB)
IBM Rational
Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 
741598. 
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU
_______________________________________________
Oslc-Core mailing list
Oslc-Core at open-services.net
http://open-services.net/mailman/listinfo/oslc-core_open-services.net