After a looong time trying to make MO work within Protege, I, well...
I did not entirely succeed. As I wrote earlier, I put every necessary
import, corrected all the non-owl-dl-ness that wonderweb owl validator
gave and pellet, and still the same error (cast exception on a RDF
untyped resource).
I noticed that the BIO vocabulary was giving the same error (and
completely crashing Pellet), and that FRBR was giving extremely weird
stuff in Protege (although it succeeds to load): almost all the
concepts/relationships are seen as individuals. So I removed the BIO
import, just to check, but apparently, we have the same sort of
problem in MO.
In despair, I tried with Protege alpha 4, and... it works... (as well
as the online version).
So I did the following changes (not committed yet, waiting for
approval from people on the list), to, at least, not make Pellet and
other owl reasoner shout too much:
* Adding some owl:imports, removing the references to the BIO
vocabulary, for now
* Typing every property as owl:[Datatype|Object]Property
* Typing every concept as owl:Class
* For datatype properties, change every range rdfs:Literal to the good xsd:...
* Replacing owl:sameAs by owl:equivalentProperty, when appropriate
The modified version is available at http://moustaki.org/resources/mo.rdf
And if anyone knows what's happening with previous versions of Protege...
Cheers,
y
With some help from the Protege people, I finally succeeded to write a version
of MO which works with Protege 3.3
It is available there:
http://moustaki.org/resources/mo.rdf
It is still far from being fully DL though (although tending towards it).
I'll commit the changes if I don't receive negative feedbacks about that.
The main problem was the namespace used for FRBR, which was wrong in
MO (using vocab.org instead
of the purl URI).
Cheers!
y