Re: Open Library and RDF

On Sun, Aug 15, 2010 at 05:52:55PM -0400, Young,Jeff (OR) wrote:
> The Semantic Web didn't make sense to me until Linked Data came along.
> As someone who sees the light and is passionate about grasping the
> opportunity, I'm worried about this "it might be wiser" assertion. UML
> is like mental floss which encourages me to assume disjoint classes.
> When I use UML to model other people's thoughts they seem to "see" that
> such conflations create serious communication/reusability problems. If
> no effort is made to systematically differentiate concepts, people never
> even realize the communication/reusability problems are being created
> and persisted. 

I do see the point in using UML to make distinctions explicit.

> The discipline and relationships between UML and OWL/RDF reveal things
> about one another and the models they represent. I wish SW experts
> understood UML better and vice versa. I hope LLD XG takes time to look
> at its conceptualizations from both perspectives.

Realistically, LLD XG does not have the charter or time to
look that closely at the conceptualizations.  In a broad-brush
way, it can merely identify areas where further work is needed and
recommend follow-on actions.

> > I take you to be saying that the official "strong" view
> > reflects a specialized view of one part of the community.
> 
> IMO, the most important point of FRBR is to establish WEMI as disjoint
> classes. The class named "Book" isn't as practical in the modern world
> as it used to be. ;-)

Okay, but how _formally_ does that disjointness need to be
expressed?  Does it need to be declared formally, so that a
reasoner will actually raise an exception when a resource is
found to be a member of two disjoint classes?  Or would it
be enough to express that disjointness informally, so that
catalogers simply know that a resource is supposed to be one
or the other, but not both?  Even informal rules can be tested.
Must the disjointness be hard-wired into the vocabulary?

I'm not suggesting that I know the answer, but I feel confident
suggesting that the question is worth close scrutiny.
Such questions came up alot in the design of SKOS, and the
group generally came down on the side of light specification.

> >  Do they not anticipate that
> > commercial companies, government agencies, or other types of
> > organizations might want to embrace FRBR concepts and adapt
> > them to their needs?  Raising the bar ontologically would
> > help ensure that their use is limited to controlled contexts
> > (or at any rate used incorrectly outside those contexts).
> > Or is that perhaps the point?
> 
> I agree that the FRBR family needs to be defined carefully in OWL to
> avoid becoming a semantic backwater. I wish it was clearer why this is
> important.

+1!

Tom

-- 
Thomas Baker <tbaker@tbaker.de>

Received on Sunday, 15 August 2010 23:08:29 UTC