RE: OWL Full reasoning

Hi Danny,
 
In closing some responses (I'll repeat your points, so that you (and others)
don’t have to search).
 
Thanks for your help!
 
Regards,
Hans
 
--------------------
 
[DA] The naming clash may be confusing things here, e.g. owl:Class is the
ClassOfClass in OWL, but almost certainly doesn't mean the same thing
as ClassOfClass in ISO 15926. So rdfs:subClassOf  *might* adequately
express the "Valid Connection per  ANSI B16.5" relation in the ISO
15926 model. (I don't know ;-)
[HT] People conversant in 15926 lingo will not have that problem. In an open
world environment like OWL you will see many homonyms, so you'd better get
used to the idea.
 
[DA] I'll get on mine ;-)
XML validation will give you syntax checking, but your example
suggests you want more than that. Using RDF will give a shareable data
model (in the sense that it could potentially be shareable across
industries/domains, not just between companies in the same domain).
[HT] The use of RDF and OWL does give shareable data thanks to ontology
merging (which unfortunately gives an exponential growth in
combinations).and proper reasoning. By putting an ontology based on a
rigorous data model like ISO 15926-2 plus a Reference Data ontology on top,
and then merging user ontologies to it, we create the ISO standardized
rigour that is required when you want to exchange and integrate data over
space (the umpty facilities in the industry) and time (the lifetime of those
facilities).
 
[DA] What you'd have with the XML Schema representation is a shareable
syntax, but the data model would remain dependent on your current
modeling language. Which is a good question - what is your current
modelling language? How are the logical relations expressed? To what
extent can you do inference across it?
[HT] Using XML Schema was a mistake in our case, we move on to OWL. The
modelling language of ISO 15926-2 is EXPRESS (defined in ISO 10303-11). You
can find the model in EXPRESS on:
http://www.infowebml.ws/ECM4.5/ECM4.5.html 
Since the ISO 15926-2 data model has the entire lifespan of facilities as
scope, we didn't include rules in it (rules often are application
domain-specific).
 
 [HT] I figured that we will need to write code ourselves indeed. That still
leaves me with the question: What's the use of OWL? 
[DA]Provides a common data model with a useful level of logical
constraint. In a case like this you may need stuff outside of RDF/OWL,
but at least there's a good foundation.
[HT]Agreed.
 
[HT] You can find the Ontology for Data Model under: 
http://www.infowebml.ws/links/ontology-for-data-model.rdf 
[DA] Thanks.
[HT] You're welcome! By the way, I corrected some mistakes, so when you
click now, you get the corrected version.
 
[DA] Sounds like it's a job for Protege [3]. It's a powerful IDE which
includes OWL support and a variety of reasoners. One caveat - in my
experience it's a little flaky at times (crashing on certain imports
for me). I don't know, if you have your data model fully expressed in
another language (entity-relation or something) then there may be a
direct way of translating across - Protege supports allsorts through
plugins.
[HT] I'll say nothing negative abour Protégé, but I have problems in working
with it. But I will study their 118-page user manual again.
 

Received on Friday, 19 August 2005 13:22:41 UTC