Re: Small problem with Expression.owl

Bijan Parsia wrote:
> On Jun 23, 2005, at 11:27 AM, David Martin wrote:
> 
>> Bijan Parsia wrote:
> 
> [snip]
> 
>>> But we're not really using SWLR, so what's the diff?
>>
>>
>> Sure, we are using SWRL.
> 
> 
> Not really. AFAIK, we aren't using any SWRL rules and we aren't using 
> the SWRL semantics *per se*. We're ripping off some of the 
> syntax...which is perhaps the most fluid bit.

We have arranged for service preconditions, effects, and 
control-construct conditions to be specified using the RDF-based syntax 
(by and large, see below).   We have one or more tools that support the 
user in constructing these expressions.  Some of us are designing and 
planning to build systems that load these expressions into a SWRL 
reasoner and check the satisfaction of the expressions in connection 
with enacting and reasoning about services (and for all I know, some 
folks may have already built such systems). So to me it's quite 
reasonable to say that we are using SWRL.

I'm aware of the issues.  E.g., we've stretched things a bit by using 
AtomLists outside of rules.  (But by and large we are using the syntax 
in accord with the spec., and the way we are going beyond that is 
straightforward.)   The semantics talks about what it means for an atom 
to be satisfied and I  believe it requires only a straightforward tweak 
to a SWRL reasoner (or possibly no tweak at all, depending on how the 
reasoner is designed) to test the satisfaction of an AtomList 
independently of any rule.  There is some unhappiness about including 
the unquoted SWRL expressions in our OWL files, but that particular 
debate isn't about whether or not we are *using* SWRL.

I have a feeling if DAML+OIL / OWL had never been "used" in any 
"stretched" ways during its "fluid" days a good deal of interesting work 
might never have been done with it.

Anyway, no need to debate further.  Why don't we just agree to disagree 
about what the meaning of "use" is?

Regards,
David

> 
>>> Why not encourage swrl to use it too?
>>
>>
>> Personally I wouldn't object to that, but I suspect from the point of 
>> view of the SWRL authors, it might be felt that tying in with 
>> standardly-defined terms like &rdf;#nil makes SWRL more meaningful.
> 
> 
> My belief is that the SWRL authors didn't care about the RDF syntax.
> 
> I expect that they would want the use of List to be "compiled out" the 
> way it is in OWL DL now. But then OWL-S should point this out and 
> encourage the *deletion* of all the swrl stuff before processing with an 
> OWL reasoner.
> 
> (Note, this is exactly a key reason I pushed for Literal based quoting 
> in the first place.)
> 
>>> SWRL isn't a standard...OWL is. I prefer to conform to the latter.
>>
>>
>> I'm not sure I see your point here, with respect to the current topic. 
>>  How would SWRL be conforming to OWL by using the shadow list vocab 
>> instead of refering to &rdf;#nil?
> 
> 
> Well,it will be OWL Full, so that's conforming in that sense. If we want 
> it to be OWL DL, we can't just say, 'Treat this extra stuff as like the 
> use of rdf:list in owl dl syntax'.
> 
> My point was that I don't view the SWRL syntax as being as worthy of 
> following to the letter, since it's just a note. There's more fluidity.
> 
> Cheers,
> Bijan.

Received on Thursday, 23 June 2005 22:38:02 UTC