Dan,

I can give you a definition in terms of Lbase.

Otherwise, it is really upto whoever is building the higher level language. They can put absolutely anything they want in there ...

In any case, any mechanism to "exempt" a bit from rdf can be stretched/used to exempt a lot. So, for example, in description logics, Classes and associated machinery (subClassOf, type, etc.) belong to the logic and this opening we provide can be used to keep all that hidden from rdf.

In the end, the different layers have got to want to get along ... There is no way for rdf to technically constrain the upper layers so that the whole thing hangs together coherently. Shotgun weddings usually don't work.

guha

Dan Brickley wrote:

This is a nice high level overview, thanks.

I would like to see "to deal with logical machinery" fleshed out in more
detail. Can you offer a more detailed account? What counts as "dealing
with logical machinery"? (I fear a repeat of the rdfs:ConstraintResource
mistake, where we didn't have a clear notion of which future
classes/properties would be flagged).

Dan