Search Mailing List Archives
[protege-owl] How to detect undefined references in rdf:resource
bunsz at gmx.de
Tue Jan 26 04:03:42 PST 2010
let me explain it in other words.
I'm trying to use owl-ontologies in conjunction with swrl in order to
perform a semantic verification of specific xml-files. The verification
goes beyond what is possible with an xml-schema.
So I have three ontologies:
1. a class ontology which consists of class / property definitions and
some enumeration class definitions and its individuals.
2. a rule ontology which imports the class ontology and contains swrl rules.
3. an individual ontology which imports the rule ontology and which is
generated out of an xml-file using xslt.
So in my xml-file I have for example:
<signOnStreet refId="Sign_3" pos="3.2" />
<signOnStreet refId="Sign_4" pos="2.5" />
<function name="stop" />
<function name="pedestrian_crossing_warning" />
Sign_3 is referenced and defined ... not so Sign_4 which is referenced
only, but not defined. Transforming this xml-file to the individual
ontology result in sth. like
<rdf:RDF .... >
And now I want to spot the error of the undefined reference to Sign_4.
This was an example, but in my case it will always be the same situation
of having an object property which refers to an undefined individual and
that is what I want to detect.
One can argue that I can transform the xml to another xml with xlink /
xpointer statements and spot undefined references while validating
against an xml-schema, but I really would like to know if this can be
done in owl somehow.
> Actually, Sign_4 is defined in this ontology.
Ok, the n3 notation is not very expressive regarding my issue.
Talking about ontologies ... "mentioned" is not the same as "defined",
is "mentioned" (referenced) and different from
which expresses a definition, am I right?
> It is explicitly mentioned, and in a context in which it is clear
> that it must be an Individual.
In my cases it will always be like that: Referenced but undefined
> The only way you would get something that doesn't have some implicit
> type is to have a mention of it where there is no context that allows
> figuring out if it is a class, property or individual. So you would
> need something like
> default:Something .
> with no other information. Just about anything else you might say
> about it would give some clue as to what sort of object it would need
> to be, and then it would be recognized and treated as such an object.
Picking this up ... is it possible to model following:
Every individual which is not defined (which has no base class but
owl:Thing) but referenced, can be classified as default:Something ...
Even if the reference is within an object property, where - by its range
- the reasoner can guess that the individual must be for example a Sign?
I hope someone can help me.
More information about the protege-owl