Re: Peter's slides about the MOF metamodel

On Aug 5, 2008, at 9:41 PM, Conrad Bock wrote:
> Bijan,
>>  I answered: All of it usefully generalizes, afaict, over the current
>>  concrete syntaxes.
> How about the RDF/XML syntax that can bundles everything "about" a  
> class
> under one element (eg, disjointWith, subclassOf, etc, inside one
> <owl:class></owl:class> element)?

I have always thought of them as abbreviations for the corresponding  
axioms. Contrast with the contents of an insersectionOf list.

They have alternative forms. Indeed, at the RDF level there is no  
bundling at all.

> In the OWL 2 W3C abstract syntax this
> is broken up into separate elements (one for subclassOf, one for
> disjointness, etc).

As on the triple level.

> The OWL 2 metamodel would likewise use separate
> elements for this, whereas the OWL 1 metamodel (or an OWL 2 metamodel
> based on RDF/XML) would put them under one.

Well, speaking from an RDF point of view, that seems very wrong.

>>  The current flavor. I'm genuinely confused. I can look at a diagram
>>  and see how it corresponds to the functional syntax and the xml
>>  syntax and the manchester syntax and the rdf syntax without too much
>>  difficulty.
> Sure, you do the mapping in your head.  I would say most hackers on
> RDF/XML will want something more direct.

More direct for me doesn't seem to mean what you think it means.

> The W3C abstract syntax is useful of course, but the question is for
> who.  If the W3C abstract syntax becomes more commonly used, rather  
> than
> just a favorite of DLers,

Sigh. It's not our favorite. At least not mine. I do think it more  
accurately reflects the structure of the language and in a useful way.

> then a metamodel for it would be as important
> as one based on RDF/XML.  I don't think that's happened yet.

I sincerely doubt that RDF/XML (with heavy emphasis on the XML) is  
really what most RDF people are thinking of when they think of RDF or  
OWL. And where the triples don't align neatly (e.g., when you have a  
restriction) I think it's much more helpful to think of them as a  
unit (rather than as a bunch of triples). Which is what the metamodel  
effectively does. I think it would be better for people coming from  
an rdf pov to conceive of restrictions as *expressions* and not  
"anonymous classes". Thinking of them as anonymous classes gets the  
emphasis wrong and leads to a whole lotta problems.

I stand by my statement. I think that the current metamodel is  
appropriate for all our current concrete syntaxes.


Received on Tuesday, 5 August 2008 21:13:46 UTC