If we're going with a one-to-one model between Work and Instance, why 
not a "one" model? Admittedly I am a "unitarian" when in comes to FRBR 
group1 and have long argued for a "whole" rather than a set of parts. 
Remember, bibframe is the data structure, not the cataloging concepts 
and FRBR is a conceptual model. If it works best for catalogs to think 
in terms of WEMI I'm all for it. But that doesn't mean that the data 
structure has to be divided into four parts -- the purpose of the data 
structure is to store the data in a way that you can do a whole variety 
of interesting things with it.

Even more than unitarian, I espouse the heresy that not all communities 
and applications will define WEMI in the same way (two parts, six parts, 
a gradation?), so pre-defining it in the library data could be a barrier 
to interoperability. Call a "work title" a "work title" and let others 
decide how that fits into their bibliographic view. Also, limiting 
relationships to a particular group 1 entity is, IMO, a great danger to 
our ability to interact with non-library data.  In other words, we can 
define our data elements to mean what we want them to mean, but we 
shouldn't try to control how others will use them or what they can link 
them to.


On 11/27/12 11:51 AM, Eric Miller wrote:
> Hi Diane, Philippe,
> In this draft model, the aggregation would be done at the (BIBFRAME) Work level. A set of collected stories, for example, would in fact be a Collection (Work) of other Works. Works are typed and a Collection is just one kind of Work. What connects these Works together are a set of defined relationships that create a Web of Works. The paperback (or hardback, or …) version of this collected Work would be instances of this Collection (but there may be other instances associated with the individual Works as well).
> Whats missing from this primer are more detailed use cases and example applications which help demonstrate more concretely this approach. I'm hopeful these will be made available soon. Nothing helps ground any model (and serialization, vocabulary, constraint layer, HTTP services, etc.) like practical examples.
> Diane, as you know no good idea goes unpunished ;), if there are specific examples that you would be willing to suggest that would be extremely helpful. I can't claim a quick turn around, but I'd be happy to give a go at representing these via BIBFRAME.
> --eric
> On Nov 26, 2012, at 10:34 AM, Diane Hillmann <[log in to unmask]> wrote:
>> Philippe:
>> I had the same response, immediately thinking of things like collected stories (single or multiple authors) and serial issues. These are certainly fairly common in bibliographic metadata, and were not well handled in MARC or MODS. The FRBR model, though admittedly a complex beast, accommodated these materials, and gave some hope that there could be ways of handling those kinds of materials in ways that a machine could understand, rather than (as usual) depending on the human user to figure it out.
>> Diane Hillmann
>> Sent from my iPad
>> On Nov 26, 2012, at 9:13 AM, "LE PAPE, Philippe" <[log in to unmask]> wrote:
>>> Dear Everybody,
>>> Colleagues have already expressed their concern about the vanishing of FRBR Expression entity in Zepheira's "BIBFRAME" model, but I was rather puzzled by this:
>>> "Each BIBFRAME Instance is an instance of one and,only one BIBFRAME Work." P. 10.
>>> What about (FRBR) Manifestations embodying more than one (FRBR) Work then? Will there be something like compound (BIBFRAME) instances? Or will the aggregation be done at the (BIBFRAME) Work level?
>>> Ph.
>>> -- 
>>> Philippe Le Pape
>>> Mission Normalisation
>>> 227 avenue Professeur-Jean-Louis-Viala
>>> Tél.  33 (0)4 67 54 84 67
>>> Fax  33 (0)4 67 54 84 14

Karen Coyle
[log in to unmask]
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet