Re: [xml-dev] 'is-a' Relationships in XML?

[Stephen Green <stephengreenubl@gmail.com>]

Thinking a bit more about it, I accept that the #some-xpath
might be unworkable of course (and alternatives like XPointer
not sufficiently supported perhaps) so maybe the GRDDL
XSLT named template with the template name in the RDF's
'about' URL is most promising. Either way it would seem to
be essential to have this as a 'standard' convention or at least
to give the convention some name (unless that too is catered
for in the present or future GRDDL specs to enable this use
case).

Best regards

Steve
---
Stephen D Green




On 10 May 2010 10:09, Stephen Green <stephengreenubl@gmail.com> wrote:
> But if the RDF is standalone and GRDDL has not been inserted into the
> schema, is there a way to point the RDF 'about's at markup XPaths?
> Would this work?
>
>    <rdf:RDF  rdf:about = "some-url#some-xpath"
>
> Then if this were a convention, maybe some extended SPARQL functions
> could extract the '#some-xpath' for a given 'some-url'.
>
> This would allow queries from the RDF side.
>
> But I accept that it would be great if the query could find the schema and
> from that find the RDF via GRDDL and XSLT.
>
> Does all this work well enough to warrant standards developers and other
> markup architects putting the time in to writing GRDDL XSLT templates and
> expressing their semantic model for their markup in RDFXML (or OWL)? If
> they did, would it allow query and assertion predicate expressions to be
> written using the semantic model rather than making assumptions about the
> markup structure?
>
> Best regards
>
> Steve
> ---
> Stephen D Green
>
>
>
>
> On 10 May 2010 09:01, stephengreenubl@gmail.com
> <stephengreenubl@gmail.com> wrote:
>>
>> Wow. Yes, that fits. Great.
>>
>> The same convention could be used in
>> standalone RDF in hope SPARQL expressions could be resolved to
>> retrieve results from markup. Then predicate assertions bound to
>> SPARQL could be used to test or validate markup based on ontology
>> rather than structure. Cool. There would be a need to have multiple
>> 'rdf:about' XPaths for a given standalone RDF model.
>> Then a way to associate each with a markup namespace, which could be included with a SPARQL, etc expression.
>>
>>
>> If this convention works it will need a name I guess.
>>
>> Best regards and thanks
>>
>> Stephen D Green
>>
>> -----Original Message-----
>> From: Rick Jelliffe
>> Sent:  10/05/2010 7:59:53 am
>> To: stephengreenubl@gmail.com
>> Cc: xml-dev
>> Subject:  Re: [xml-dev] 'is-a' Relationships in XML?
>>
>> stephengreenubl@gmail.com wrote:
>>> If it works, maybe GRDDL is the
>>> solution. Otherwise, maybe something similar to the OASIS
>>> context / value association file but with RDF or OWL classes or
>>> properties, etc as the 'value' would help and maybe give more
>>> granularity than GRDDL to the context. GRDDL associates RDF via
>>> XSLT with an element or type definition in the schema.
>>> Associating RDF or OWL with any context which can be expressed
>>> using XPath might be better. I'm not sure I like using XSLT to create RDF
>>> for a context. That might obfuscate the reverse lookup where from the
>>> RDF side you want to find the XPath for a given class or property or resource.
>>>
>> I suppose your generated RDF could contain a link back to the
>> xs:template, by which the XPath (and any priorities) could be
>> retrieved.  something like
>>
>> <xs:template id="T1"  match="s/n/a/f/u">
>>
>>    <rdf:RDF  rdf:about = "{concat( $GRDDL-URL, "#T1") } "   ...
>>
>> Cheers
>> Rick Jelliffe
>>
>>
>>
>>
>