Re: Tim Berners-Lee on the Semantic Web

From: Jonathan Rochkind <rochkind_at_nyob>
Date: Wed, 21 Oct 2009 15:43:33 -0400
To: NGC4LIB_at_LISTSERV.ND.EDU
All of this makes sense. BUT once someone HAS established a work for The 
Conquest of Bread and filled in it's subjects, I shouldn't _have_ to go 
assigning subjects every time I make a new manifestation record, should 
I?   I should be able to inherit from that work, such that when the work 
subjects are changed (someoen does a better job), my manifestation knows 
that too.

It needs to work both ways. And we definitely need to prepare for a 
world where different pieces of the puzzle get filled in at different 
times by different people.   I shouldn't need to know everything when 
cataloging an item in front of me, but when someoen else who does know 
more adds more, it should all go together properly. 

One hypothetical flow scenario: I make a manifestation record for 
Conquest of Bread, and I fill in subjects for it. Later someone else 
comes along and looks at our shared cataloging database, and says, gee, 
this is really a part of this work set, and a record exists for that. 
(Yes, a work is a SET, but a set can be an entity too! And can have 
attributes!).   They tie it to the work set. Maybe they're asked "You 
want to merge in the subjects from the previous free-standing 
manifestation, or discard them?"

Another workflow that's actually the same workflow, but is represented 
differently in the record structure:  I make a manifestation record for 
Conquest of Bread, and I don't care to find a pre-existing work record 
for it (maybe I don't have time, maybe one doesn't exist), but the 
system automatically creates a placeholder work record for me, and 
attaches my subjects to that. Later, that same someone else comes along 
and says "Gee, this is really a part of this other existing work set", 
and merges the two work records.

Those are actually the SAME workflows, these are system implementation 
details.  There are also workflow details to be worked out to deal with 
_expecting_ different pieces of data to be added by different people at 
different times (and 'data' includes relationships and set memberships).

But you've got to do all that on top of a formalized domain model. 
Having a formalized domain model is what even gives us the language to 
have this discussion. And the work you do on top of the formalized 
domain model will no doubt give you insights suggesting changes in the 
underlying formalized domain model, in an iterative process. That's 
fine. But you've got to start sometime, and you've got to have a 
formalized domain model. 

Which I why I find it very counter-productive to be saying that basing 
things on FRBR is a negative, and we should just go on without any 
formalized domain model at all.   Basing things on FRBR as a formalized 
model doesn't decide things one way or the other in terms of the 
implementation issues we'll have to work out to make the scenarios Karen 
talks about possible. But going with FRBR as a formalized model to begin 
with _allows_ us to start working on those implementation issues. As RDA 
was trying to do -- whether it has been succesful or not is another 
story, but that (at least starting a few years ago) it was _trying_ to 
base itself on the FRBR model is not to it's detriment.

Jonathan

Karen Coyle wrote:
> Matthew & Jonathan:
>
> I agree with both Matthew's bottom-up and Jonathan's sets. To me, a Work 
> is a grouping of Expressions, and that means that the "workness" exists 
> even if you don't have the information you need to create a Work entity. 
> (One of my problems with WEMI is that there seems to be an assumption 
> that all of the information to fill in all of the entities will be 
> there... which is not the case, see below.) Each level, to me, is the 
> sum of the levels below it, so the manifestations add up to what you 
> currently know about one or more expressions, and what you can deduce 
> about the work. It's a build-up process.
>
> My "resource" btw was intended to be more "superWork" -- WEMI, in my 
> mind, are parts of something, not wholes in themselves. The fact that 
> they are not whole is what gives me a hard time in thinking of them as 
> entities. A Manifestation doesn't have subjects? Or an author? Hmmmm. So 
> WEMI is an interdependent set of descriptions, not separate entities. 
> That gives me pause.
>
> So how will we, in practice, fill these in? I used as an example a book 
> that I found on the street in Berkeley (I get great books out of boxes 
> of discards here):
>
> The Conquest of Bread
> by Peter Kropotkin
> New York Vanguard Press1926
>
> No mention of translation. If I'm cataloging this and don't want to go 
> hunting around for more info, I cannot fill in much relating to the W. I 
> don't know the original Work title, I don't know what language it was 
> originally written in. This is probably not the first version, since he 
> died in 1921.
>
> I need to create a manifestation "entity" without a work "entity". But I 
> want my manifestation to have an author and subject headings. What can I 
> do? I think I should be able to create an entity based on what I know. 
> And if it goes into a larger context where manifestation author: Peter 
> Kropotkin and manifestation title: The conquest of bread matches a 
> record that also has a link to a work with:
>
> Author     Kropotkin, Petr Alekseevich, kni?az?, 1842-1921.      
> Work Title     Conquête du pain.
>
> Then my data can hook up to that data and in a sense gets connected to 
> the W, which I couldn't do on my own. But maybe I gave it the subject 
> "Agriculture and Communism" and the W it hooks up to gave it the subject 
> "Anarchism". Then my subject enhances the W.
>
> So the W is an ever-changing set that is the sum of the E's and M's (and 
> I's) in its universe. And if it's designed as linked data, all of these 
> parts can more easily come together.
>
> [In my mind, which draws much better than my hands do, these are like 
> tinker toys or jigsaw puzzle pieces, or even molecules, that float 
> around and fit together when they meet. So I may have piece f, g, and l 
> and the Work has piece g, s and z, and when they meet they form g, f, l, 
> s, z.]
>
> kd
>
> -
> -----------------------------------
> Karen Coyle / Digital Library Consultant
> kcoyle@kcoyle.net http://www.kcoyle.net
> ph.: 510-540-7596   skype: kcoylenet
> fx.: 510-848-3913
> mo.: 510-435-8234
> ------------------------------------
>
>   
Received on Wed Oct 21 2009 - 15:46:19 EDT