Re: owl.owl

At 17:13 +0100 3/11/03, Guus Schreiber wrote:
>I have heard some say that they expect people to explicitly import 
>the owl.owl file in any OWL ontology. I'm not sure this is the right 
>strategy. However, if we want poepl to do this there is a problem: 
>owl:Nothing is defined in owl.owl, so OWL Lite ontologies importing 
>owl.owl would automatically become OWL DL ontologies. [Also, the 
>Dublin Core annotations need to be typed with 
>owl:AnnotationProperty, but that can be fixed.]

Owl.owl is more deeply problematic than that.

In OWL Lite and OWL DL it is simply not allowed to add annotations to 
owl:Class or owl:Ontology.

My take is that owl.owl is part of the RDF compatibility requirement, which is 
addressed by OWL Full and semantic layering.

Hence, we should only be aiming to support the importing of owl.owl by OWL 
Full ontologies.

In OWL Full it is legal, and possibly even encouraged, to define your own 
subclasses of owl:Class etc. and so Jim's concerns appear ill-founded to me.

We probably need clarity that owl.owl is *not* automatically imported by OWL 
Full - since the annotations cause hiccups. Particularly those on owl:Thing 
and owl:Nothing which would be legal in OWL DL, and hence would problematize 
semantic layering if they were necessarily true (in the formal sense).

Jeremy

Received on Wednesday, 12 March 2003 08:17:26 UTC