Re: Some questions on properties

On Thu, 27 Apr 2000, Stefan Decker wrote:
>>Two new questions.  First looks like a bug in the 02/22 specs:
>>
>>================================================================
>><?xml version="1.0"?>
>>
>><!-- Example from REC-rdf-syntax, demonstrating abbrevation -->
>>
>><rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
>>          xmlns:s="http://description.org/schema/"
>>          xmlns:v="http://description.org/view/">
>>   <rdf:Description about="http://www.w3.org/Home/Lassila">
>>     <s:Creator>
>>       <!-- REC-rdf-syntax just says `about' -->
>>(*)   <s:Person rdf:about="http://www.w3.org/staffId/85740">
>>         <v:Name>Ora Lassila</v:Name>
>>         <v:Email>lassila@w3.org</v:Email>
>>       </s:Person>
>>     </s:Creator>
>>   </rdf:Description>
>></rdf:RDF>
>>================================================================
>>
>>In the (*) line, it says simply `about', but if I understand things
>>correctly this brings `about' in the `s' names space, resulting in
>>a quite different interpretation.  Right?
>
>"about" and "ID" are somewhat not affected by namespaces, when used as 
>defined in the
>RDF-Specification.

Hmm. I don't like too many `somewhat's' in a specification.  Could you
clarify?  Does it mean that, if an attribute has no explicit
namespace and the name can be interpreted as RDF we should do so?
This makes sense, but my XML parser adds all namespaces (if I ask
it to respect namespaces) and I have no way of telling how the
namespace got there (explicit or by default).

Now, of course I can change that, but `it just doesn't feel right'.

>>Second is again on the protertyElt production [6.12].
>>
>>What is the difference between
>>
>>   <rdf:Description ID="foo">
>>     <a1 ID="bar">
>>        <rdf:Description>
>>           <name>Hello World</name>
>>        </rdf:Description>
>>     </a1>
>>   </rdf:Description>
>>
>>and
>>
>>   <rdf:Description ID="foo">
>>     <a1>
>>        <rdf:Description ID="bar">
>>           <name>Hello World</name>
>>        </rdf:Description>
>>     </a1>
>>   </rdf:Description>
>>
>>I have the impression they should be the same and specifying both ID's
>>inconsistent is an error.  Right?
>
>The first definition gives the ID "bar" to the statement {foo,a1,genid1}
>the second definition gives an ID the second resource and leads to the 
>triple {foo,a1,bar}.

I see.  That makes sense.  Should the following be regarded a proper
translation (both SiRPAC, and the Pro Solutions compiler only give the
obvious three triples).

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#"
         xmlns:s="http://description.org/schema/">
  <rdf:Description about="http://www.swi.psy.uva.nl/projects/SWI-Prolog/">
    <s:Creator rdf:ID="pl-creator">
      <rdf:Description ID="JW">
        <name>Jan Wielemaker</name>
        <works_at>SWI</works_at>
      </rdf:Description>
    </s:Creator>
  </rdf:Description>
</rdf:RDF>


{'JW', name, literal('Jan Wielemaker')}
{'JW', works_at, literal('SWI')}
{'http://www.swi.psy.uva.nl/projects/SWI-Prolog/', 'http://description.org/schema/':'Creator', 'JW'}
{'pl-creator', rdf:type, rdf:'Bag'}
{'Statement__1', rdf:type, rdf:'Statement'}
{'Statement__1', rdf:subject, 'JW'}
{'Statement__1', rdf:predicate, name}
{'Statement__1', rdf:object, literal('Jan Wielemaker')}
{'pl-creator', rdf:'_1', 'Statement__1'}
{'Statement__2', rdf:type, rdf:'Statement'}
{'Statement__2', rdf:subject, 'JW'}
{'Statement__2', rdf:predicate, works_at}
{'Statement__2', rdf:object, literal('SWI')}
{'pl-creator', rdf:'_2', 'Statement__2'}
{'http://www.swi.psy.uva.nl/projects/SWI-Prolog/', 'http://description.org/schema/':'Creator', 'pl-creator'}

	Regards --- Jan

P.s.	I've started a test-suite, now containing 37 examples with
	their desired output, mostly from the spec document.  Anyone
	with tricky (but preferably short) rdf examples?

	Ideal would be the definition of an XML format for the triples,
	so any parser can use the suite for validation.  Generated id's
	complicate this a little.

	As soon as the parser and test-suite have matured enough it will
	be made available under the LGPL (test-suite public domain).

	People interested in testing may contact me.

Received on Thursday, 27 April 2000 05:03:25 UTC