On Thu, Oct 22, 2009 at 09:42, Jonathan Rochkind <rochkind_at_jhu.edu> wrote:
> I just still don't understand your concerns on FRBR itself. It seems pretty
> good to me. It seems a lot better than nothing.
Hmm. No, a bad model can be quite worse than nothing. With nothing, at
least you're not in the negative, which you will do if you go too far
down the wrong path. It's not like you've got unlimited tries and time
to take the next step.
> It was based on formalizing
> 100 years of cataloging practice, and I think this is the right way to go.
Hmm. I think I just got a slight allergic reaction to the reasoning
here. 100 years of formalized cataloging practice is *not* a good
thing when a lot of people think a 100 years of cataloging practice is
partly / wholly to blame for the mess we're in right now. I think I
understand what you're trying to say (that we shouldn't scrap all
we've built up), but the onslaught of electronic resources surely will
have a dramatic (if not cataclysmic) effect on the model we should
use.
> Sure, it needs some tweaks, perhaps a whole lot of tweaks, but it's
> fundamental basis seems sound to me.
Can anyone with more formal FRBR ties explain the process for change
or tweaking? Or if at all possible? I can understand even if it is
not; FRBR has lied dormant for more than 10 years before the recent
flurry of interest.
> What is the actual alternative to going ahead with FRBR? Another 10 years
> of dickering around and not going forward obviously isn't a solution. You
> and I (if not some) agree that tinkering with AACR2 as rules for creating
> narrative text without an underlying formalized model isn't a solution
> (although not everyone in this discussion does). So... what?
Ok, let's talk alternatives. I'm afraid I might get a bit technical
here, but as far as I'm concerned, the FRBR best feature is to take an
important step out of record keeping and into basic epistemology,
specifically representialism. However, I'm very much with Karen on the
strange distinctions between entities and properties and how FRBR
seems to be in a world of their own, but communication and a bit of
tweaking could work that out. (However, this isn't the first time this
has been raised)
One of those things wrong is identity as an entity (unless they're
knee-deep in proxy identification handling, and then they've got the
problem of the relationship between Identities and Things for multiple
proxies, which is another story). For any reasonable identity
management (people who know me knows how much I love to bang on about
this topic :) to happen, for practical reasons and for not forgetting
that the RDF framework is your probably best ally, it is done through
proxy, meaning a set of properties that infer identity. RDF does this
through the notion of non-discrete sets of URIs (while Topic Maps have
subtle distinctions of identity, where the proxy lives inside the
Topic itself and have three directions of inference) which you must
govern with tools, which is a bit of a bummer but a cold reality.
Very quickly we run into the epistemological problem of Twain. Is the
identity "Mark Twain" and "Samuel Clemens" identical? Is there a
relationship between the two, or are they the same? And what happens
when we add the lesser known "Thomas Jefferson Snodgrass" to the mix?
(Feel free to try to solve the puzzle)
For bibliophiles these are inherently important puzzles to solve, and
the first thing with FRBR that strikes me is that it doesn't even try
to handle this, or, if it does, I simply haven't found it. (Please
enlighten me!) Looking stuff up in WikiPedia and sorting it all out is
a very human endevour, but how to model it? How to model a works
importance in coherence with the identity of the writer, fictional or
not? Is Lemony Snicket an identity or an entity? Are we sure?
So, what are alternatives? Well, has anyone thought of creating a
basic ontology for the world (Thing -> Living thing -> Animal ->
Primate -> Human, etc. with a whisk of relationships defined in broad
terms), and let it be extendable, and let the finality of certain
things be worked out over time? The tools are there to do it, if you
choose wisely. (And I say this most humbly; the situation was quite
different only 5 years ago) May I suggest not a model which we need to
sort out, but rather a simple set of guidelines for creating an
extendable ontology, and through that get better answers down the line
without loosing time right now to get started? There's too much at
risk in getting a rigid model like FRBR up and running. I think
Manifestation and Item are both fine (although I have some
reservations on Manifestation), but Work and Expression are subtle,
dangerous and fickle things. We still talk about whether this thing is
this or the other, a sure sign that these things aren't as settled as
they should be, 15 years later.
Anyway, a few bobs worth.
Regards,
Alex
--
Project Wrangler, SOA, Information Alchemist, UX, RESTafarian, Topic Maps
--- http://shelter.nu/blog/ ----------------------------------------------
------------------ http://www.google.com/profiles/alexander.johannesen ---
Received on Wed Oct 21 2009 - 23:11:44 EDT