Re: Disjointedness of FRBR classes

On Sun, Oct 30, 2011 at 12:08:47PM -0400, Jon Phipps wrote:
> Karen, I don't believe any of the current FRBR models 

By "models", do you mean formal representations of the models in RDF, e.g.:
    http://iflastandards.info/ns/fr/frbr/frbrer/frbrer.rdf - FRBRer
    http://www.cidoc-crm.org/rdfs/FRBR1.0.1.rdfs - FRBRoo
    http://vocab.org/frbr/core.rdf  - Ian Davis

Or do you mean the texts on which representations are based, e.g.:
    http://www.ifla.org/VII/s13/frbr/frbr.pdf = http://www.ifla.org/VII/s13/frbr/frbr.htm - FRBRer
    http://archive.ifla.org/VII/s13/wgfrbr/FRBRoo_V9.1_PR.pdf - FRBRoo
    http://vocab.org/frbr/core.html - Ian Davis
    -- based on http://archive.ifla.org/VII/s13/frbr/frbr.htm ??

>                                                       support this notion
> of 'statement bundles' (distinct from WEMI entities) very well, and it's
> not clear to me at the moment how best to model it in RDFS/OWL, although I
> think Tom's suggestion to use 'named graphs' in some way is worthwhile. 

If Named Graphs were defined formally -- and I expect they will be in the next
release of RDF -- then I picture a particular WEMI entity, for example an
instance of an Expression, as a named set (or "bundle") of statements, one
statement of which might be that the described resource is a instance of the
class Expression.

The individual statements in that bundle would all be about the described
resource.  However, the "bundle" (named graph) itself would also be related to
the described resource, perhaps via a predicate that means (in essence)
"describes".

If the idea of saying a given resource is a Work and then -- without formal
contradiction -- saying it is also an Expression, is considered too strange to
contemplate, one could perhaps drop the idea of assigning identity to Works and
Expressions directly and, instead, type the Named Graphs themselves along WEMI
lines.

For example, a given Named Graph could be declared to be an instance of a
WorkDescription.  A WorkDescription would be a description of something from a
"work" point of view.  Might such an approach be of some practical benefit
while avoiding the (unintended?) costs of WEMI-as-disjoint-classes?

> I'm not convinced that FRBR properties represented as 'shareable units' is
> workable in the Open World context of Linked Data and is liable to be more
> useful in system-local metadata creation/maintenance strategies rather than
> distribution/aggregation.

I think you mean: "I'm not convinced that FRBR _entities_..." ??

Named graphs would be not only shareable, but referenceable, so instead of
locally replicating statements which, let's say, describe the resource as a
Work, one could just link to a named bundle of such statements.  I understand
this to be be one of the key potential benefits of using FRBR.

Tom

-- 
Tom Baker <tom@tombaker.org>

Received on Sunday, 30 October 2011 19:08:04 UTC