Franck, hello.
On 2008 May 3, at 16:28, Franck Le Petit wrote:
> We are building a vocabulary that will be used in the Simulation
> Datamodel. Semantics is required for several "keywords":
> - Physical processes
> - Algorithms
> - Simulated Object
> - ...
> May you tell me what is the best way to implement this ? Should we
> do .RDF files ? And if yes, is there any tool to do it easily, I
> mean to facilitate the entries and the relationships between words ?
The semantics group is currently discussing a document which aims to give precisely this sort of advice.
The current Working Draft document is at
http://www.ivoa.net/Documents/WD/Semantics/vocabularies-20080320.html
and the current "editors' draft" lives at
http://www.astro.gla.ac.uk/users/norman/ivoa/vocabularies/
(I've just today released a fairly significant update to that). Summarised, it says that yes, RDF is good for this, and that specifically there's a framework called SKOS, which is intended to support precisely this sort of vocabulary. I'd be very keen to hear your comments on the intelligibility and user-friendliness of this document, if you have time.
Vocabularies are at the lightweight end of the ontology spectrum, and SKOS vocabularies in particular are concerned with rather thin semantics such as which terms are broader or narrower in scope than others. The sort of ontologies that Ed is talking about are more expressive -- you can do more with them -- but they're correspondingly more expensive to create and broker agreement on.
The most practical approach, I think, is to find a vocabulary that is reasonably close to what you need, and use that as much as possible. Creating a vocabulary (or ontology, or schema, or DTD) is easy; creating one that a useful number of other people agree on is very hard!
The vocabularies that are distributed with the above WD (a snapshot of which you can download as a bundle at <http://www.astro.gla.ac.uk/users/norman/ivoa/vocabularies/rdf/vocabularies-2008-05-08.tar.gz >) were generally produced by starting with an existing vocabulary in some semi-regular format, and writing a per-vocabulary script to regularise that into RDF. That is, we didn't actually need to use a GUI tool at any point, though I'd agree with Ed's remarks about the appropriateness of Protégé.
Best wishes,
Norman
-- Norman Gray : http://nxg.me.uk Physics and Astronomy, University of LeicesterReceived on 2008-05-08Z20:09:51