Re: swrl.owl is OWL Full and Protege does not support OWL/XML

From: nvdesai@ncsu.edu
Subject: Re: swrl.owl is OWL Full and Protege does not support OWL/XML
Date: Thu, 14 Oct 2004 18:46:20 -0400 (EDT)

> > I remain puzzled as to how swrl.owl even plays into the Jena parsing
> > situation.  It is not as if swrl.owl defines the semantics or even the
> > syntax of SWRL.
> 
> [I use SWRLX to mean the OWL/XML syntax of the language i.e. swrlx.xsd and
> SWRL to mean OWL/RDF syntax of the language i.e. swrl.owl. I also use SWRL
> to mean the semantic web rule language.]
> 
> Peter,
> 
> I think one more note is worth the confusion :-). Maybe, I am not asking
> an intellectual question.
> 
> I am not worried about whether SWRL or SWRLX defines the syntax or
> semantics of SWRL. But one would take one of them as their starting point
> for defining owl-rule ontologies. So, my concern is on the practice side
> of the matter.
> 
> Root of the problem is that any development needs tool support (yes,
> graphics are good). And all editing tools use Jena for parsing/spitting
> purposes. And Jena is an RDF parser !!
> 
> swrl.owl allows me to use Protege, either instantiate or subClass SWRL
> classes, and develop my ontologies and rules efficiently. Which is fine as
> long as swrl.owl is not going to be deprecated and is going to be
> maintained with the development of the rule language.

I still don't see why the presence/absence/form of swrl.owl should affect a 
tool (like Protege) that can process SWRL.  Again, it is not like swrl.owl
defines the syntax of SWRL, so I don't see why/how Protege could possibly
depend on it.

> If I choose OWLX, I have to hand-write everything. But apparently the
> spec. encourages me to use OWLX.

Well, yes, to some extent at least.  RDF is a very poor language for
expressing syntax - there are just too many parts of the SWRL syntax that
don't naturally fit.  

> Converters are of no use if SWRLX is well developed while its SWRL
> counterpart is not. They basically do not allow me to leverage goodies and
> refinements of SWRLX as they will lose that in conversion.
> 
> For example, swrl.owl will allow any rdf:resource in argument1 and
> argument2 of individualPropertyAtom, regardless of the domain and range of
> the respectuve objectProperty/DataTypeProperty. I am not sure, but I
> suspect this would not be the case with SWRLX.

Again, swrl.owl has no normative impact.  

> If usability of SWRL is not a significant point of worry at this stage,
> then there is no point in my question. If Jena supports OWLX in near
> future (sounds very unlikely given their RDF centered model), and Protege
> makes updates to use new Jena model, then also there is no problem.
> Or else, if swrl.owl is refined, maintained, and given equal importance
> (which is unlikely as you mentioned deprecation of RDF), then also there
> is no problem.
> 
> SWRLX and OWLX are good but how would one use them if tools are not going
> to support them ?
> 
> Excuse me for a lengthy post,
> 
> -Nirmit

I remain puzzled as to what role you expect swrl.owl to play here.  Note
that swrl.owl is described in http://www.w3.org/Submission/2004/03/ as
[an] "OWL ontology partially describing the RDF Concrete Syntax of SWRL".
	           ^^^^^^^^^ ^^^^^^^^^^ 

Peter F. Patel-Schneider
Bell Labs Research

Received on Thursday, 14 October 2004 23:35:22 UTC