Re: exchanging OWL through RIF

On 3 Mar 2006, at 12:24, Francois Bry wrote:

>
> Dear All,
>
> Obviously, the following usages can be expected, because some need  
> them:
>
> 1. translating OWL formulas/specifications in a RIF.

in my understanding, for OWL, we will need (as for any other "formalism"
we might want to consider) a translator that takes some OWL and maps
it into RIF. It is, however, not the job of RIF to provide this  
translator.

Moreover, there could be different such translators: e.g., one  
translator could simply extract the
(implicit) class hierarchy from an OWL ontology, whereas others could  
take
ALL of this ontology --- in case that RIF captures full first order  
logic (the real one,
with existentials in the head and a possibly infinite domain).


> 2. querying OWL ontologies in (the antecedent of) a RIF rule - what  
> was
> refered to as "theory reasoning" two days ago.


ok, this is what I called the "uni-directional" approach.


> 3. using RIDF rules for extending/complementing OWL specifications.

ok, this would be my "bi-directional" approach.

> In my opinion, we should try to keep all these usages possible. 1
> and 2
> seem to me much easier than 3 and therefore preferable as Phase 1  
> goals.
> (My understanding of Uli's mail is that she would favour 3 for  
> Phase 1).

hm, for phase 1, I favour an approach where (3) is well-understood and
powerful, i.e., what I call the  bi-directional approach.

> Making 1 possible

As indicated above, I agree with "making 1 possible" but
not with "Realizing 1"

Cheers, Uli

> seems to me to be essential, because otherwise, as Ed
> (Barkmeyer) pointed out, the providers of rule software would not get
> the help from a RIF theyt need coping with OWL.
>
> Regards,
>
> François
>

Received on Friday, 3 March 2006 12:52:16 UTC