On Dec 6, 2007 9:08 AM, Jonathan Rochkind <rochkind_at_jhu.edu> wrote:
> But the real answer is: Whatever is the most convenient way for us to
> model it.
But that is - when it comes down to it - the real question I'm asking.
Also, I don't think we should base the future of catalogin on what's
convenient as much as what makes sense and feels right.
> I still think that the FRBR gives us a fine model, it's a skeleton on
> which we can hang our data in which ever way is most convenient (ideally
> in standard community decided upon ways---and that we haven't spent the
> last 10 years thinking about these ways is a pity)----but it's useful
> that we're all at least using the same skeleton.
Well, the first problem is of course that we're not using it. A
selected few crazies are experimenting with a prototype of it,
perhaps, but *we* as a community are certainly not using it, and we
are definitely not all understanding the concepts the same. The
discussions here of late more than proves the ambiguity of the current
model, and maybe it's an education problem more than anything, but my
gut feeling is that if we need *training* to understand a basic
conceptual model, we're screwed.
The second problem is the notion of FRBR as a skeleton on which to
build on. It's not. Either you agree with or understand the basic
conceptual model, or you don't, and if you don't your meta data will
not only be a poor match when mixed with other data, but might spoil
the completeness of your data pool. (It's one of the big hidden perils
of ontology work :)
The third problem is more of a problem statement, like you say
"ideally in standard community decided upon ways." This is the crux of
the whole FRBR success thing. I for one do not believe that community
driven 'agreements' are enough for ontology work. Something either is,
or it is not, and you need to be strict and specific about it, and it
needs to be written down, printed in blood and sealed with a thousand
seals (and so forth and so on). I'm of course talking about a model
such as FRBR which has a base ontology you cannot pick and choose from
; you need to embrace the basic model fully, or it will not work
(unless you live in a cave). Of course, if there was an ontology level
*above* FRBR that has community support and flexible relationships
built in, then no problem, but I doubt very much we will be able to
pull that one off ; we can't even agree on the basics of FRBR. (Not
that we *can't*, but there certainly is much confusion)
> Useful for
> inter-community community, useful for explaining what we are doing to
> others (they can learn the skeleton, and then learn how we have hung
> things on it), useful to ourselves in in understanding what we are doing
> clearly, and in giving our various standards a common point of reference
> (that skeleton).
Hmm. I don't share your enthusiasm for and wonderfulness of the powers
of RDF modeling, I'm afraid. Semantic data modeling in an ontological
fashion is extremely hard to do for experts, so how hard will it be to
grasp for mere mortals? I'm not saying it's not worth the effort, and
with the right tooling you can get a long way, but just because it's
modeled doesn't mean people can use the model. Of course, all the
things you say above are true ; these things *can* happen, provided
all librarians now become ontology work experts with the right tools
to boot. What are the chances of that?
> FRBR isn't perfect, but it's good enough to start with, and a hell of a
> lot better than nothing.
Possibly. Hey, don't get me wrong, I've argued for FRBR for years, so
it's not really the technology I've got a problem with. I've got a
problem with seeing librarians grasp this stuff fast enough for us to
do something worthwhile with it. 1) grasp the model 2) grasp semantic
data modeling 3) grasp persistent identification 4) grasp tools for 2)
and 3). I'm not saying it can't be done, but I don't find it very
realistic.
To have most people (in our case, librarians) grasp a model, that
model must be very simple. The more complex the model, the more the
expert you need to be in the model to get it all. People can grasp
Dublin Core, but people cannot grasp MARC (and no, I don't include
catalogers in the 'most people' bucket; they're too special for that
:)
And I'm more than willing to say that FRBR is too complex for quick
uptake. Not sure if it can be trimmed down, but I suspect it just
might. (Maybe someone should have a stab at it, reproducing FRBR on
two pages so we all can grasp it)
Anyways, yay! to FRBR, but not so yay! to the ambiguity of it.
Regards,
Alex
--
---------------------------------------------------------------------------
Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps
------------------------------------------ http://shelter.nu/blog/ --------
Received on Wed Dec 05 2007 - 18:19:07 EST