That's an interesting idea, but the concept "view" suggests that there
is an underlying... um, thing, that you are _viewing_. There might be
different views, but they are different views of the same underlying
thing. If this is so, that underlying thing has to be a metadata
representation too, doesn't it? This doesn't to me seem to get around
the fundamental problem here.
To put it another way---do you think that one "view" might make
_different_ decisions about Work-Entity-Manifestation relationships than
other views? If so... how could we possibly manage to control all that,
we have enough trouble just trying to control _one_ set of decisions
about those things. And if so, that means that each "view" lives in
it's own seperate universe, they aren't really just "views" of the same
universe at all, they're entirely seperate universes, and if you have a
case where you need to bridge those universes, you're going to have
trouble, no? But if not, and each "view" makes the same decisions about
these things, they just _expose_ different aspects of the underlying
decision (this is what is connoted by the very term 'view', right?),
then underlying all those "views" is still a single metadata and domain
model with a single set of decisions, we're back at the problem of
trying to design a data set that works for everything---even if that
dataset is _exposed_ in different "views".
Jonathan
Riley, Jenn wrote:
>> So the response from some is now that we need to design the
>> metadata for some particular new application, but we need to
>> design the metadata to flexibly support many (any?)
>> applications. Are we wrong to be thinking this? Is this a
>> sisiphean task? Do we need to specify the bounds on what
>> sorts of applications we mean to support better (the FRBR
>> user tasks are one attempt to do this; are they sufficient?
>> Why or why not?).
>> Should we specify the bounds all the way down to something as
>> specific as one particular application? (The 'online public
>> access catalog'?
>> Surely not!)
>>
>
> Sarah Shreeves of UIUC and I have been promoting recently the idea of
> "shareable metadata" - that which is intelligible and useful in metadata
> aggregations. We start from the notion that Carl Lagoze championed in
> the early days of DC that all metadata is a *view* of a resource - that
> it isn't possible to create a usage-neutral metadata record. Our
> thinking here is that we don't have to design a view for every single
> application, but that we can present a different view for every major
> *class* of applications - a Google-style view, and OAIster/OAI-PMH style
> view, a collection registry-style view, etc. I think we have a ways to
> go before we fully understand what those fundamental views are and what
> metadata they should contain, but it seems to me that a happy medium
> between designing for every single application and designing for one
> single application is the way we should be thinking about this.
>
> Jenn
>
> ========================
> Jenn Riley
> Metadata Librarian
> Digital Library Program
> Indiana University - Bloomington
> Wells Library W501
> (812) 856-5759
> www.dlib.indiana.edu
>
> Inquiring Librarian blog: www.inquiringlibrarian.blogspot.com
>
>
--
Jonathan Rochkind
Sr. Programmer/Analyst
The Sheridan Libraries
Johns Hopkins University
410.516.8886
rochkind (at) jhu.edu
Received on Wed May 23 2007 - 07:47:31 EDT