[Pellet-users] Inference problem

[Bijan Parsia]

On Sep 28, 2007, at 8:56 AM, Role Francois wrote:

> (1) > The understanding SWRL series may help
>>    http://clarkparsia.com/weblog/category/semweb/rules/swrl/
>
> Thanks. I have looked at these interesting posts. Here is my  
> understanding
> at the moment. Please correct me if I am wrong.
>
> In the example in part 2, :mary and :sheevah  will not be found to be
> instances of :AncestorOfACreep for the following reason :
>
> If we interpret the rule below  in DL Safe mode
>
>  ancestorOf(?X, ?Z) :- ancestorOf(?X, ?Y), ancestorOf(?Y, ?Z).
>
> then ancestorOf(?X, ?Y) cand "bind" to ancestorOf(mary,akane) or
> ancestorOf(sheevah,akane) but ?Z can not bind to anything because  
> no creep
> individual has been explicitly named.

Yep.

> Therefore :mary and :sheevah can not
> be deduced to be instances of :AncestorOfACreep.

Yep.

> (2) >  Yes I wanted to mimic subsomption using rules.
>> Note that the rules are DL Safe, so you won't infer a subsumption  
>> relation
>> between classes.
>
> I used the phrase "mimic subsolmption" because I thougt the rule to be
> similar to stating   Boy subClassOf Person. Is that a wrong  
> interpretation
> ?

The DL Safe rule is weaker than the class axiom. For example, if you  
have
	Boy subClassOf Person
	Person subClassOf Animal.

You would expect to conclude
	Boy subClassOf Animal.

But if you take the DL safe versions...you won't. (Except in very  
special cases.)

So it only mimics subsumption to a certain extent. The SWRL version  
of the rule would have exactly the same consequences as the class axiom.

Cheers,
Bijan.