Re: LANG: syntactic version for imports (and other things)

I am really missing something, and I still haven't heard an answer on 
this:

WHAT IS THE POINT of referencing a symbol from another external ontology 
if you don't get, as a consequence of referencing it, the meaning of that 
symbol as defined in that ontology? 

As others have already repeatedly pointed out, there is no way to 
"extract" the definition of a symbol from an ontology, so the only way to 
get it is to include the whole ontology it is defined in.

This has nothing to do with reasoning over the entire web.  It's just 
reasoning over the part of the web you claim you are consistent with.

-Chris

Dr. Christopher A. Welty, Knowledge Structures Group
IBM Watson Research Center, 19 Skyline Dr.
Hawthorne, NY  10532     USA 
Voice: +1 914.784.7055,  IBM T/L: 863.7055
Fax: +1 914.784.6078, Email: welty@us.ibm.com




Frank van Harmelen <Frank.van.Harmelen@cs.vu.nl>
Sent by: www-webont-wg-request@w3.org
09/25/2002 10:54 AM
Please respond to Frank van Harmelen

 
        To:     www-webont-wg@w3.org
        cc: 
        Subject:        Re: LANG: syntactic version for imports (and other things)

 




Bernard Horan wrote:

> If I wish to point a reasoner at
> document 3 to ensure that (say) it has no inconsistencies (in the same
> way that I can point FaCT at an OilEd document), does this mean that I
> have to indicate somewhere that the ontology in document 3 should
> 'import' the ontology in document 2? Should it also 'import' the
> ontology in document 1? Or should it only 'import' the ontology in
> document 1 if that ontology is not imported by the ontology in document
> 2? I.e. what's the transitive character of 'import' here? And how do I,
> as a user, know which ontologies I should be importing??

I have always assumed that when drawing any inferences from any OWL 
statements, you will have to state from which OWL statements you want to 
draw 
the inferences.

The obvious ways would be to point at a particular <owl:ontology>.
This would then include any imported ontologies, but >*not*< any of the 
ontologies from which classes/properties are used without explicit import 
statement (but simply by mentioning them).

I would expect toolbuilders to provide you with the option to explicitly 
mention that you want these additional ontologies to be taken into account 
as 
well, but this would require an action on your part, it would not be 
automatic simply because someone uses a URL from another ontology.

Both our semantics doc and our guide doc should make clear that inferences 

are drawn with respect to a given set of premisses, and also how these 
premisses are obtained (by pointing to an ontology (or: several), and 
thereby 
getting all its imports as well; not by assuming that every relevant 
resource 
on the web is somehow magically included).

This is also my standard answer to the perenial question about 
inconsistency 
on the Semantic Web: I won't reason with respect to the entire Web, but 
only 
with respect to a set of premisses that >*I*< indicate.

Frank.
   ----

Received on Wednesday, 25 September 2002 15:10:14 UTC