-This is an interesting point. Using a DL reasoner for this type of checking is certainly an option. Yeah, cerntainly not the way it's meant to be used, and a coarse way of doing model checking. But it can be an easy, somewhat effective built-in "secondary indicator" of a problem with your model. As for checking, the one that has gotten me is assiging a subject to a datatype property that's of the incorect "range" (e.g. using an int instead of a string). A pre-defined or auto-generated ruleset that does this built in type checking for you (so I wouldn't have to write all the constraints myself for each particular model) would be great. Maybe even warnings for assignment of things that are only satisfiably of the specified range (or specified restriction) class of a property, to let you know of the specific classing inference that is going to occur, for models where you plan to assert your individuals' classes. Maybe call it the Strong Typing Ruleset. Or is that too much "closed-world" thinking?
________________________________ From: Holger Knublauch [mailto:[email protected]] Sent: Monday, January 26, 2009 1:44 PM To: [email protected] Subject: [tbc-users] Re: Annotation Properties - With OWL-DL ontology/data, as our ontology changes, and affects the pre-existing data, OWL-Full data usually indicates some inconsistences--sometimes I have outdated references to classes/properties that no longer exist, for example. This is an interesting point. Using a DL reasoner for this type of checking is certainly an option. Another option would be to use some kind of a rule and pattern based approach that detects such "errors" in your model. For example, the eyeball tool from the HP Labs (Jena group) might help. Or, a collection of SPIN rules encoding best practices might be helpful. We are actually working on such a rule set. It would be great to get some details on which things should be checked. - untyped resources as predicates - untyped resources as rdf:type object - what else? - When TBC indicates OWL-Full, it puts the offending data into the Error Log, implying that OWL-Full is an error. :) Well this is certainly just for our own laziness - we could have created an extra view for those warnings. However, once we have the SPIN rules in place, it will be more convenient to simply browse them as warnings in the Problems view. But the second part of the question was, when rdfs:label is an AnnotationProperty with domain/range, why is this *not* tagged as OWL-Full, when it (I think) violates the spec? (I don't *want* it to be tagged, I just want to understand why it is not). Are these hard-coded in the reasoners, since if I create my own annotation property with a range, I move into OWL-Full? I believe the system triples are ignored altogether by species checker because they would take any OWL DL model into OWL Full anyway. Holger --~--~---------~--~----~------------~-------~--~----~ You received this message because you are subscribed to the Google Groups "TopBraid Composer Users" group. To post to this group, send email to [email protected] To unsubscribe from this group, send email to [email protected] For more options, visit this group at http://groups.google.com/group/topbraid-composer-users?hl=en -~----------~----~----~----~------~----~------~--~---
