W3C home > Mailing lists > Public > public-owl-comments@w3.org > July 2009

More background on OWL dot OWL file

From: Holger Knublauch <holger@topquadrant.com>
Date: Tue, 21 Jul 2009 09:39:31 -0700
Message-Id: <F056ED5E-459E-48B5-9FF9-AF3337A99CD1@topquadrant.com>
To: public-owl-comments@w3.org
Thanks for responses to my request [1] for an owl.owl file of the OWL  
2 vocabulary. I have seen some comments [2], [3] on the other mailing  
list, and I'd like to clarify one of the main reasons for my request.  
As I am not a group member, I need to use this comment mailing list  
here.

In TopBraid, we actually "import" the system triples into the union  
graph of each model that the user opens. This makes it possible for  
the tool to use generic algorithms in many places and reduce the need  
for hard-coding specific vocabulary terms.

For example, the following triples (from Michael's draft)

	owl:ReflexiveProperty rdf:type rdfs:Class .
	owl:ReflexiveProperty rdfs:subClassOf rdf:Property .

make it very easy for the tool to enable users to create a new  
reflexive property: We simply display a generic class selection dialog  
with a tree rooted at rdf:Property so that users can pick the  
metaclass they want to instantiate. Both triples above are needed for  
this to happen. And I would strongly suggest to also have rdfs:labels  
and rdfs:comments on all those terms, which would then be displayed  
(automatically) as tool tip texts, or when the user navigates to the  
definition of the owl:ReflexiveProperty class.

There are numerous other places where having the "metadata" of the  
system classes and properties available, to drive the user interface,  
to traverse class trees in (SPARQL) queries, and simply to have  
consistent algorithms that significantly reduce the burden on the  
programmer. I am confident that many other tool vendors or people  
implementing linked data browsers will share this view. And, having  
the OWL triples in the same union graph is no problem for other  
algorithms that don't need them - for example we do not send those  
system triples as input to some inference engines which already have  
them hard-coded in their engine.

In a nutshell, having an owl.owl file makes it possible to treat OWL  
almost like any other vocabulary such as SWRL. This is a very good  
thing in the spirit of the evolution of the Semantic Web, in which OWL  
is one possible vocabulary among others.

Thanks
Holger

[1] http://lists.w3.org/Archives/Public/public-owl-comments/2009Jul/0007.html
[2] http://lists.w3.org/Archives/Public/public-owl-wg/2009Jul/0030.html
[3] http://lists.w3.org/Archives/Public/public-owl-wg/2009Jul/0032.html
Received on Tuesday, 21 July 2009 16:40:15 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Tuesday, 21 July 2009 16:40:15 GMT