RE: Test ontologies

These sorts of things are extremely useful for our testing and 
implementation experience (important for CR).  I would like to find a 
better way than simply sharing on this list to get these out there -- 
We could consider adding these as tests in the test suite 
(particularly the Galen, as we have the "answers" as well) - or, we 
could consider putting these somewhere visible on the web site and 
letting people access and use to do testing without a more formal 
structure.  I believe the latter might be worth considering - during 
the CR period we could do some of this both to help implementors and 
to gain momentum for our eventual PR request.  At one point I had 
talked about the possibility of creating http://www.w3.org/2003/OWL 
as a web site and putting some useful links there (like the RDF site 
did) - with the idea that the WG would start this, and then transfer 
it to whatever interest group was started in the Semantic Web 
Activity at W3C.   If people think this is a good idea, I think we 
could do several useful things on such a site -- point to some of the 
"idioms" we've discussed, provide some examples like those below, etc.
  -JH
p.s. Ian, if you haven't already, why don't you submit those to the 
http://www.daml.org/ontologies library?  It now accepts OWL.



At 1:47 PM +0100 7/25/03, Ian Horrocks wrote:
>On July 25, Peter Crowther writes:
>>
>>  > From: Ian Horrocks [mailto:horrocks@cs.man.ac.uk]
>>  > I converted some "interesting" ontologies to OWL:
>>  >
>>  > http://www.cs.man.ac.uk/~horrocks/OWL/Ontologies/ka.owl
>>  > http://www.cs.man.ac.uk/~horrocks/OWL/Ontologies/galen.owl
>>  > http://www.cs.man.ac.uk/~horrocks/OWL/Ontologies/mad_cows.owl
>>  >
>>  > They have been checked with both validators and seem to be OK. The
>>  > first (ka.owl) is Lite, the other two are DL. "galen.owl" is (part of)
>>  > the well known Galen medical terminology ontology - classifying this
>>  > ontology should prove an interesting challenge for new reasoner
>>  > implementations.
>>
>>  In particular (speaking from experience), there are some
>>  less-than-obvious subsumption relationships in galen.owl - and it
>>  doesn't even use several of the more interesting features of DL.  What
>>  Ian (characteristically) hasn't said is that this fragment of the Galen
>>  KB is something of a benchmark for DL reasoners; the fastest
>>  implementation of a description logic (SHF, not OWL DL) of which I am
>>  aware can calculate the full subsumption hierarchy of galen.owl in a
>>  hair under 10 seconds on a 2GHz Pentium, excluding the file load time.
>>
>>  Ian, do you also have a list of expected subsumptions in galen.owl?  It
>>  would provide a useful target for DL reasoner implementations.
>
>I have attached a dump of the taxonomy of named classes from the
>ontology. It is set of lists of the form:
>
>(class-name direct-supers direct-subs)
>
>where class-name is the name of a class, direct-supers is a list of
>its direct superclasses, and direct-subs is a list of its direct
>subclasses. E.g.:
>
>("BodySpace" ("Space" "BodyStructure") ("PericardialSpace" 
>"OperationField" "BodyHole"))
>
>says that the direct superclasses of "BodySpace" are "Space" and
>"BodyStructure", and that the direct subclasses of "BodySpace" are
>"PericardialSpace", "OperationField" and "BodyHole".
>
>Regards, Ian
>
>
>>
>>  		- Peter
>>  --
>>  Peter Crowther, Director, Melandra Limited

-- 
Professor James Hendler				  hendler@cs.umd.edu
Director, Semantic Web and Agent Technologies	  301-405-2696
Maryland Information and Network Dynamics Lab.	  301-405-6707 (Fax)
Univ of Maryland, College Park, MD 20742	  *** 240-277-3388 (Cell)
http://www.cs.umd.edu/users/hendler      *** NOTE CHANGED CELL NUMBER ***

Received on Friday, 25 July 2003 09:00:48 UTC