Re: Peter's slides about the MOF metamodel

On 31 Jul 2008, at 16:59, Conrad Bock wrote:
> Can you identify the portions of the current metamodel that would be
> common across all the concrete syntaxes available?

I don't see any part that's *not*. Where do you see it not being?

We use the conceptual model *everywhere*. Entities, expressions,  
axioms... this is a *HUGELY* good thing. (Various of us spent time  
distilling this model.) One thing I think it's *vital* when coming to  
OWL from an RDF perspective, for example, is that one should shift  
one's mental model way from triples. Triples are *not* the basic  
expressive unit of OWL in the way they are in RDF. Some triples are,  
themselves, axioms. Others are parts of expressions.

Sometimes you can fruitfully ignore that. But most of the time you  
can't, in my experience. Even when you are thinking all tripley it's  
important to understand the difference between:
	:p rdf:type owl:TransitiveProperty.
	:p rdf:type :Person.
and even
	:p rdf:type owl:ObjectProperty.

These have *critically* different roles (even in OWL Full; of course,  
in OWL Full there's *also* a role they all have in common; frankly,  
that's horribly confusing).

>> If we proliferate metamodels, then we raise the cost of concrete
>> syntaxes considerably.
> Fewer metamodels is better than more,

One is best.

> and of course we should continue
> trying to do that, but it's better to have more if important user  
> groups
> refuse to use the few we think they should.

Yes, if sufficiently important user groups rebel at what we do enough  
we should change what we do or point them to where others are doing  
what they want to do. Is that the case here?


Received on Thursday, 31 July 2008 16:14:34 UTC