Vocab AND Ontology?

[Bernard Vatant]

Tony, and all

On the OWL vs SKOS debate, I would like to quickly stress again a point 
I tried to make before
http://www.ivoa.net/forum/semantics/0709/0362.htm
... but which was maybe blurred by other points in this verbose message 
;-) .

SKOS is not a "low-level-semantics" OWL, because SKOS and OWL have not 
the same functional aims. Again, re-read above message for more details.
Choosing either one of those representations, or both, should not be 
based on feasibility/complexity of representation, but on functional 
objectives. Let me take an example illustrated in both Ed's Star 
ontology and IAU Thesaurus, the "White Dwarf Star" concept
- If you are a documentalist want to index/search/retrieve 
documents/data about "White Dwarf Stars", a SKOS concept "White Dwarf 
Stars" is what you need, and the position of this concept in a Concept 
Scheme hierarchy, and its association with other concepts such as 
"Chandrasekhar Limit".
- If you want to classify objects, find them using logical inference, an 
OWL class "White Dwarf Star" is what you need, with its necessary 
conditions, attached properties, position in a subsumption tree etc.

If you want both uses, you need both representations, BUT : *you can't 
expect to automatically and safely transform one representation into the 
other automatically*.
- From SKOS to OWL : not every concept defined in a SKOS concept scheme, 
and used for indexing/search/retrieval makes sense as a class in a OWL 
ontology. See the "cosmology example" in my previous message.
- From OWL to SKOS : only concepts represented as classes would be 
available for indexing, so you would have to add indexing concepts which 
were not present as OW classes.

Look at White Dwarf in IAU Thesaurus and in star ontology. In the former 
you find the related concept "Chandrasekhar Limit", which makes perfect 
sense for indexing, because there is a lot of litterature for which 
Chandrasekar limit is a relevant indexing concept, and even the primary 
topic. But you don't find "Chandrasekhar Limit" at all in the star 
ontology, although it's a very pertinent concept in stellar 
astrophysics, but there again it's perfectly normal because this concept 
does not fit in as a class, nor even as an individual or property value. 
One can imagine a formal definition of the white darf class including 
indirectly this concept by putting a constraint on the mass (although 
this is tricky in OWL), but you get the picture.

This example again to illustrate why I'm concerned about any methodology 
based on : let's do SKOS first, and then we will transform into OWL, or 
the other way round.

Bernard

Tony Linde a écrit :
>> Correspondingly, the OWL is
>> trivially produced from either format if you're interested (some of
>> us are definitely NOT).
>>     
>
> This seems the core question, Rick (whatever the starting point and I agree
> that the IAU Thesaurus seems a good one from my naive pov): do we produce
> the SV in OWL and derive the SKOS or produce it in SKOS and derive the OWL?
>
> I'd still plump for the former for the following reasons:
>
> a. either approach seems, at the outset, to involve little cost (though I'd
> still assert that the availability of Protege and other ontology tools make
> this a cheaper route);
>
> b. any extensions to the ontology will break the link between vocabulary and
> ontology since the extensions could not be derived from the vocab so we
> could well end up with divergent terms meaning the same thing;
>
> c. it seems that no matter how the ontology (with vocab included) is
> extended, we can still derive the vocabulary providing we mandate that vocab
> terms use the classes and relationships specified.
>
> This is a fairly simplistic view, I know. Can anyone show any methodological
> or technical problem with this approach (I don't think Rick pointed out any
> real problems with the OWL-based approach: what ontological baggage might
> cause problems)?
>
> And this is still only my view - others will have their own position on the
> question above.
>
> T.
>
>   
>> -----Original Message-----
>> From: owner-semantics at eso.org [mailto:owner-semantics at eso.org] On
>> Behalf Of Frederic V. Hessman
>> Sent: 24 September 2007 10:14
>> To: semantics at ivoa.net
>> Subject: Re: Vocab AND Ontology?
>> ...
>>     
>
>
>
>   

-- 

*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/>