Search Mailing List Archives


Limit search to: Subject & Body Subject Author
Sort by: Reverse Sort
Limit to: All This Week Last Week This Month Last Month
Select Date Range     through    

[protege-owl] instances inferring

Thomas Russ tar at ISI.EDU
Thu Apr 26 12:02:03 PDT 2007


On Apr 26, 2007, at 9:19 AM, Abrami Géraldine wrote:

> As far as I understood the Open World Assumption, when I create a  
> new “empty” class (without restrictions) MyNewClass directly under  
> owl:thing and run the reasonner I thought that all individuals  
> would have been classified in this MyNewClass.
>
> Why does this not happen?
Because there is no reason to support them being members of the class.

By having a "empty" class without any restrictions, you have created  
a primitive class.  In the absence of sufficient conditions, there is  
no way that any reasoner can infer membership in the class.  The only  
way to be a class member is to be asserted as a member.  [Note,  
assertion to a subclass also works, since that is a sufficient  
condition for membership in the parent.]

The open world assumption doesn't have anything to do with this.   
What open world means is that one cannot use the absence of  
information to prove a negation.  It applies to constructs like NOT  
(Complement) or ONLY (AllValuesFrom) and Max cardinality.  Open world  
doesn't increase the set of things that can be inferred, rather it  
decreases it.







More information about the protege-owl mailing list