Re: Non- and Partial-FRBR Metadata

Quoting Ross Singer <ross.singer@talis.com>:


> So basically the question is:
>
> 1) You can figure out from the metadata elements what entity they
> should belong to
> -or-
> 2) You can't.
>
> If it's #1, then we're in good shape.  The URL
> "http://openlibrary.org/books/OL12558007M" could potentially return:
>
> http://openlibrary.org/books/OL12558007M#manifestation
> http://openlibrary.org/books/OL12558007M#expression
> http://openlibrary.org/works/OL15326163W#work (for consistency) -
> http://openlibrary.org/works/OL15326163W would also return this
>
> Given the fact that in RDF you don't know if you have all of the
> facts, anyway, it gives you the freedom to assert what you know: it
> doesn't matter if it's the entire story.  Your record also doesn't
> have to be the resource.

I'm  trying to grasp which of our problems this solves. Are you  
assuming that each bibliographic property is explicitly associated  
with one (and only one) of these URIs? I do like the idea of using  
cool uris rather than minting arbitrary URIs for the Group1 entities,  
but that seems to be a convenience rather than a fundamentally  
different structuring of the data.

One of the issues we face is that FRBR appears to require an E to link  
a W and an M. In the case where you have a smattering of data but not  
a complete description (as described in FRBR), would this allow you to  
link an M to a W even if you have no E properties? In that case, what  
would the E URI resolve to?

I think this needs an example, and I suspect that a picture would work  
better than more email text. I'll try to mock something up.

Also, I'm really puzzled by:

> Your record also doesn't
> have to be the resource.

and maybe an example will clarify that.

kc


-- 
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet

Received on Wednesday, 29 September 2010 15:35:58 UTC