Re: Open Library and RDF

On Sat, Aug 14, 2010 at 10:05:44PM +0100, gordon@gordondunsire.com wrote:
> FRBR will almost certainly declare the classes Work, Expression, Manifestation,
> and Item as disjoint. All associated WEMI properties will have a domain of one
> of these classes. The RDA model is based on FRBR, so a logical contradiction
> will result if the subject of instance triple associated with a Work is the same
> as the subject of instance triple associated with an Expression, etc.

Okay, so the classes are (almost certainly) explicitly being
declared as disjoint, and the FRBR maintainers do consider it
a logical contradiction if a resource is said to be a member
of two different classes...

Is everyone involved in the process happy with this?  It is an
ontologically "strong" view of WEMI, which seems optimistic.
With such a conceptually sophisticated model, I'd expect it
to be well understood by relatively few people and therefore
applied imperfectly in practice.  I'm wondering whether, in
a linked-data environment, data consumers might need to be
more tolerant of logical contradictions.  I guess my instinct
would have been to err on the side of under-specification, 
at least initially.

Is there a sense that FRBR should be promoted for wide uptake
or that, in practice, its use will be limited to controlled
environments with university-trained experts?  And is there a
sense that the WEMI classes are so well-defined that experts
can be expected to distinguish between them accurately and
consistently?

Tom

Received on Saturday, 14 August 2010 21:42:56 UTC