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

[Stephen Green <stephengreenubl@gmail.com>]

I forgot to add:

Maybe, to enable this, there is a way to write the XPaths into the ontology
such that the semantic query resolves to an XPath (or XQuery?) and when
this XPath in turn is evaulated it returns the answer to the semantic query.

Also assertions can be written too using semantic query expressions which
evaluate initially to an XPath (or XQuery or in-memory equivalent maybe for
improved performance) and the XPath then evaluates to true or false.

Nice thought? Seems to me to be where it leads if we do better than prose
specifications of the semantics of the markup to give us machine executability.

Stephen D Green

On 6 May 2010 13:28, Stephen Green <stephengreenubl@gmail.com> wrote:
> I'd propose some best practice principles then
>
> 1. semantic queries regarding XML should ideally be written in a semantic
> query language
>
> 2. semantic queries using XPath should be considered second best
>
> 3. semantic conformance rules / criteria should be tested using
> semantic assertions
> (just as we can test structural criteria using XPath assertions such
> as Schematron,
> XSD 1.1, test assertion markup with XPathsm and other associations between rules
> and XPaths, etc)
>
> 4. semantics can be defined for a markup using semantic / ontology languages
>
> 5. semantic expressions can be evaluated over the markup using semantic queries
> but for this, maybe further work is needed to establish how to
> associate XPaths with
> the semantic definitions
>
> Best regards
>
> Stephen D Green
>
>
>> Apologies for harping back on the
>>  former thread.
>>
>> Given then that I need to define the
>> semantics of a markup outside of that
>> markup and outside of a schema,
>> perhaps using RDF or the like, I'd probably
>> want to associate many classes and properties
>> or expressions with XPaths in my markup.
>> If I used RDF, it seems to like URIs or URLs
>> or the like (there seem to be many such
>> identifier standards of late). Would there
>> be a standard way to add something like
>> an XPath to perhaps the end of such a URL,
>> so I can point a semantic expression like
>> class or property to a node in the markup?
>> XQuery? It would be nice to have a semantic
>> query like SPARQL be able to resolve
>>  it.
>>
>> Thanks
>>
>> Stephen D Green
>>
>>
>