Search Mailing List Archives
[protege-discussion] Protégé OWL issues
tar at ISI.EDU
Fri Nov 24 20:12:50 PST 2006
This should be on the protege-owl mailing list instead.
On Nov 22, 2006, at 5:48 PM, Morozov, Anna ((DWLBC)) wrote:
> Hi all,
> We are using the Protégé 3.2.
> We have come across the following issues:
> · If a Relation is defined (without Range or Domain being
> specified) and then subsequently used in association with a
> specific Class (complete with a specification of the Range to which
> the operator some applies) erroneous data can be entered when
> specifying an Instance. In this case, an Instance of ANY Class in
> the ontology can be specified, not just those in the range
> specified when the Relation was applied.
This is correct behavior.
It would even be correct behavior if you specified range and domain
for the relation.
The important thing to remember about OWL is that range and domain
are INFERENCES and not type restrictions. The effect of making the
assertion that you note is to allow OWL to conclude that the instance
in question belongs to the range concept.
> · Complex NESTED specifications do not seem to be possible and
> consequently, when entering details for a specific Instance (when
> the parent and parent's parent details are ideally echoed as the
> nominated Relation Instance is selected) the data entry can easily
> go astray.
I don't understand what you are saying here.
More information about the protege-discussion