Ben and Chris,
I'll throw my thoughts into the mix...
On Mon, Nov 30, 2009 at 4:20 PM, Chris Wilper <[email protected]> wrote:
> Hi Ben,
>
> On Mon, Nov 30, 2009 at 6:30 PM, Benjamin Armintor <[email protected]>
> wrote:
> > http://fedora-commons.org/jira/browse/FCREPO-539
> >
> > I'm drafting an ontology, and have a few questions for the developer
> community:
>
> I would be interested to hear others' opinions on this also. Here are
> mine.
>
> > 1. How important is an OWL ontology versus an RDF schema (eg, the
> > Dublin Core scheme)?
>
> Owl seems cleaner. And it has the convenient breakdown of
> less-expressive to most-expressive versions.
>
I recommend starting first with RDFS and only going into OWL if need be.
The goal being to be able to provide at a bare minimum, a set of the
simplest definitions for the Entities in Fedora, their basic properties and
relationships to one another. Then would I begin to build up on top of this
with OWL constraints.
Look to the work done in BIBO for real world applications of OWL and Entity
modeling.
Be wary of encoding explicit "hasPart" relations, on large collections they
can grow unmanageable for real systems (use search or inference where
appropriate to resolve these).
> 2. If OWL is specifically desired, what flavor should be the target?
> > I'm inclined to dismiss OWL (full) out of hand. DL has some support
> > (cf. http://jena.sourceforge.net/inference/#owl). Lite will put some
> > severe constraints on what the schema can assert.
>
> Lite would be my preference, if it can be done, since RDF expressed in
> owl lite is easier to reason about.
>
> But if you want to get into treating fedora-model:hasModel as a
> sub-property of rdf:type, that's automatically OWL Full, if I remember
> right.
>
> IMO, the important first step is to get the basic object/datastream
> properties in machine readable form.
>
And why I recommend first approaching expressing at least a RDFS form.
>
> > 3. Should the schema describe only the core system relationships (see
> >
> http://fedora-commons.org/confluence/display/FCR30/Triples+in+the+Resource+Index
> > ), or should it include the object properties with URIs in the
> > fedora-model namespace? If the latter, should all the properties the
> > fcrepo code knows about be included (see
> >
> http://fedora-commons.svn.sourceforge.net/viewvc/fedora-commons/fedora/trunk/fcrepo-common/src/main/java/fedora/common/rdf/FedoraModelNamespace.java?view=log
> > )? That would drag some datastream properties into the mix.
>
> I think having them all would be helpful.
>
sounds reasonable.
Cheers,
Mark
--
Mark R. Diggory
Head of U.S. Operations - @mire
http://www.atmire.com - Institutional Repository Solutions
http://www.togather.eu - Before getting together, get t...@ther
------------------------------------------------------------------------------
Join us December 9, 2009 for the Red Hat Virtual Experience,
a free event focused on virtualization and cloud computing.
Attend in-depth sessions from your desk. Your couch. Anywhere.
http://p.sf.net/sfu/redhat-sfdev2dev
_______________________________________________
Fedora-commons-developers mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/fedora-commons-developers
