Vocab AND Ontology?

[Bernard Vatant]

Ed

Ed Shaya a écrit :
> Bernard,
>     Well, I have to agree to the point that OWL and SKOS have 
> different aims.  The aim of OWL is nothing less than expressing 
> knowledge in computer readable form.  SKOS is aimed at relating 
> vocabulary terms, which one has to admit is a subset of knowledge.
> Therefore, you can easily incorporate the skos terms into an OWL 
> ontology if you like.  I provide an OWL skos ontology below.
> So now you can have within OWL skos:related, broader, narrower, and 
> the various types of notes (description, historyNote, etc...), Concept 
> and subjectOf property. And keep rigorous rdf:subClass.  And make use 
> of all of the OWL tools and utilities.  In fact most of these 
> "concepts" (including Concept) were already in our ontology 
> http://archive.astro.umd.edu/ont/index.html at the top level, 
> Science.owl), but I am happy to hand these over to the skos namespace 
> as they are not really science terms.
Yes, of course. Including SKOS vocabularies in OWL classes is a 
straightforward task. Just import a SKOS vocabulary in SWOOP and it 
makes it for you :-)
>
> One should be careful though to use subClassOf if the thing is really 
> a subClass and narrower if it is not a subClassOf but a related term 
> within the scope of the broader term.
Beware that this kind of recommandation could be misleading. Be careful 
to represent something as a skos:Concept if you want to use it this way, 
and as a owl:Class if you need a cllass (I won't repeat myself too much 
...). Then use consistent realtionships, subclassOf for classes, and 
BT-NT for concepts.

> As for the specific example of white dwarf and Chandrasekhar limit 
> (CL).  In OWL you can restrict the WhiteDwarf to have Mass hasValue 
> (maxValue hasValue CL).  The CL is an instance of MassLimit.  
> MassLimit are a subClassOf  Mass.  Measurements have property sci:of, 
> which allows restriction: "CL of WhiteDwarf".

Maybe this works. I would like to see the formal OWL code, though 
(please reply privately, I don't think the list needs to see that gory 
things). Having Mass as a class with individual instances sounds weird 
... OK. I don't want to go into a detailed modeling discussion, but I'm 
afraid that before you achieve widespread understanding, let alone 
consensus, on such representation for all possible astronomical 
concepts, you'll be a very old man, and will have pile up such complex a 
model that no one will ever dare use it in real life. :-)

> The question then is, do you bother to mention that CL and WhiteDwarf 
> are also SKOS:related?

Because I will not index my docs/data with an instance of the "Mass" 
class! I will index with a simple Thesaurus concept, simply "related to" 
the white dwarf concept, without needing a reasoner to figure this 
relationship out of a convoluted restriction of the maximal value of 
some property on some class!
I will need other, simpler representations of the same "natural" concepts.

-- 

*Bernard Vatant
*Knowledge Engineering
----------------------------------------------------
*Mondeca**
*3, cité Nollez 75018 Paris France
Web:    www.mondeca.com <http://www.mondeca.com>
----------------------------------------------------
Tel:       +33 (0) 871 488 459
Mail:     bernard.vatant at mondeca.com <mailto:bernard.vatant at mondeca.com>
Blog:    Leçons de Choses <http://mondeca.wordpress.com/>