Vocab AND Ontology?

[Frederic V. Hessman]

> I'd like to propose, if it is feasible, a way forward on the issue  
> of creating an SV. Namely that:
>
> 	1. subject to discussion and resolution of the points raised in  
> Andrea's Note, sections 3.1/2/3, we proceed with the creation of a  
> SV by a subgroup of this wg;
> 	2. that the SV should be created in the OWL format, for which  
> there are several tools, not least, Protege;
> 	3. that all initial relationships between SV terms be members of  
> either the NarrowerThan or BroaderThan classes (with the inverse  
> relationship being a member of the other);
> 	4. every release of the OWL format be accompanied by a translation  
> into the SKOS format for use as a W3C standard vocabulary (other  
> formats for individuals or subgroups to create from either of these  
> releases).
> 	5. the subgroup produces a first draft within one month of the  
> interop using the terms already defined in Andrea's Note so that  
> the wg can check it out and finally sign off that this is a viable  
> way forward;
> 	6. during that same month, the wg discuss and agree on the issues  
> in points 3.1/2/3 and any others in the Note;
> 	7. when the first draft is agreed, the SV subgroup continue its  
> work, producing further drafts no slower than one per month;
> 	8. using these drafts, people who wish to work with vocabularies  
> produce test applications for discussion here and presentation at  
> the next interop (using the SKOS or other derived formats);
> 	9. using these drafts, people who wish to work with ontologies,  
> extend the SV ontology as necessary and produce applications that  
> demonstrate the use and usefulness of ontologies to astronomy, also  
> for presentation at the next interop;
> 	10. and that following these presentations at the next interop,  
> the wg decide whether to continue with a separate SV development or  
> to accept that all applications can be met with an ontology or some  
> other outcome.
> The above would ensure that the SV could be created as quickly as  
> some people seem to think is necessary using a well-known and  
> popular tool (Protege). This initial effort would have as its scope  
> only the terms useful in an SV (need to define what this is). The  
> relationship between terms would be limited to a small number of  
> relationship types, all of which must be classifiable (in  
> vocabulary terms) as narrower or broader. This latter will ensure  
> that a classic SKOS-type (and any other type) of vocabulary could  
> be derived from the ontology and that it will conform to the multi- 
> hierarchy form explained by Bernard.
Bravo, Tony, a good start.  However.....

The understandable restriction to use only NarrowThan and BroaderThan  
(the maximum amount of ontological info that could/should be kept  
within what is just supposed to be a "vocabulary"), I'd say that we  
should use SKOS for starts, which has practically no unnecessary  
ontological baggage used or implied (!) and - as far as I can see  
(but I'd be happy to be corrected) - will be more compact: assuming  
we'll want 1) the token itself (only for computers), 2) a human- 
readable equivalent, 3) any aliases, 4) any NarrowerThan's and  
BroaderThan's, e.g.

	<skos:Concept rdf:about="BARRED_SPIRAL_GALAXIES">
		<skos:prefLabel>barred spiral galaxies</skos:prefLabel>
		<skos:altLabel>SB</skos:altLabel>
		<dc:description>Spiral galaxies having a radial bar-like structure  
in their spiral pattern.</dc:description>
		<skos:broader rdf:resource="#SPIRAL_GALAXIES"/>
	</skos:Concept>

That's it (well, add more altLabel's, broader's and narrowers, and  
maybe dc: documentation about changes, but otherwise....).  This is  
exactly what SKOS is intended to do,  so unless someone can show that  
OWL is shorter/simpler.....   Even if SKOS is changed, we're just  
talking about the minimum info

	token=BARRED_SPIRAL_GALAXIES
	label=barred spiral galaxies
	alt=SB
	description=Spiral galaxies having ....
	bt=SPIRAL_GALAXIES

which we could publish as a text file.  Correspondingly, the OWL is  
trivially produced from either format if you're interested (some of  
us are definitely NOT).

------------------------------------------
The existance of ontological tools like Protoge is NOT an argument  
for OWL - indeed, is rather a counter-argument.  I don't need any  
fancy tools to put a vocabulary (e.g. SV or IAU thesaurus) trivially  
into one of the above formats and _nothing_else_should_be_added!
------------------------------------------

Secondly, I suggest we use the IAU thesaurus as a starting point :  
sure, it has lots of historical baggage and a heavy old-fashioned  
stellar emphasis, but the VO may want to keep the possibility of  
referring to old photoelectric photometry or scanned photographic  
plates etc. since it doesn't cost anything and the thesaurus has the  
_very_substantial_ political advantage of being already officially  
accepted - we want to make this process as short and painless as  
possible!   The problem with the IAU thesaurus was that it was just a  
long list with no obvious use: the VO and the HTN provide the  
framework for being able to use it.   We can look at the differences  
between the IAU thesaurus and the proposed SV to see what needs to be  
updated (a process I've already started). As a by-product, we get an  
automatic version in French, German, Spanish, and Italian - a good  
example of a secondary vocabulary document which can be published and  
immediatelly used by, e.g., the AOIM people or HTN projects which  
have to interact with the public.  The corresponding conversion  
documents for AOIM, A&A, HOU, ADS, SIMBAD,.... would follow easily,  
so that we could start with a reasonable sample of useful  
vocabularies which approaching the IVOA for final acceptance.

By keeping the vocabularies just vocabularies, the ontological work  
thereafter can be thought of as a further process by which the  
multiple vocabularies are studied in their native habitats, from  
which further ontological connections can be gleaned.   If you just  
need tokens, you just throw away a few lines of XML or use the NT's  
and BT's to help track down potentially useful alternative tokens and  
you're finished; if you want ontologies, the conversion to OWL is  
equally trivial and you can start building upon the vocabulary as  
much as you want.   Nobody can really complain.   We'd be welcome to  
publish official version as HTML (handy for determining what tokens  
are already there, especially if a few search tools are provided),  
SKOS (minimum defining XML version), and a derived OWL document for  
those who are interested.

I'm at a conference this week but will post my first cut of an  
improved IAU thesaurus for discussion next week.

Rick
------------------------------------------------------------------------ 
------------------------
Dr. Frederic V. Hessman     Hessman at Astro.physik.Uni-Goettingen.DE
Institut für Astrophysik          Tel.  +49-551-39-5052
Friedrich-Hund-Platz 1         Fax +49-551-39-5043
37077 Goettingen                 Room F04-133
http://www.Astro.physik.Uni-Goettingen.de/~hessman
------------------------------------------------------------------------ 
-------------------------
MONET: a MOnitoring NEtwork of Telescopes
http://monet.Uni-Goettingen.de
------------------------------------------------------------------------ 
-------------------------