Search Mailing List Archives
[protege-owl] OWL question about asserting individuals
tar at ISI.EDU
Thu Apr 19 08:49:20 PDT 2007
On Apr 18, 2007, at 9:11 PM, Skeptic 2000 wrote:
> Hi, I'm new to this mailing-list, so I'm not sure if it's the kind
> of questions that can be asked here.
> Let's say I have created a class named A, one named B, and X an
> object property (range and domain is irrelevant).
> A is then defined with a necessary and sufficient condition: X
> someValueOf B.
> Now I can create an individual from Thing and assert it of being
> member of A without caring of this restriction. I'm not sure to
> understand, why fundamentally, individuals asserted types can
> circumvent class sets of restrictions.
Because the restrictions sanction inference, and OWL has open world
When you assert that your individual is of type A, then the system
knows that there must be at least one filler of the X property of
type B on your new individual. But the system doesn't know exactly
what individual this filler is. But with open world semantics, the
system isn't bothered by not knowing everything.
Consider what would happen if I told you "Jane is the parent of a
son." Are you bothered by the fact that I haven't identified Jane's
son? No. But you can still answer the question "Does Jane have a
son?". It's the same way with OWL.
It is important to realize that restrictions in OWL are not like
types in programming language. The restrictions are not just
constraints that are checked, but rather they are logical assertions
that allow inference to take place. You will only get a restriction
violation if you make assertions that result in a logical
inconsistency. For example, if you were to assert that the
individual was both a member of A and a member of a (possibly
anonymous) class with X max 0. In that case, there is no way to
satisfy both restrictions, so a conflict has arisen.
In your case, I think the main item is the open world semantics.
More information about the protege-owl