Re: Tim Berners-Lee on the Semantic Web

From: Dan Matei <dan_at_nyob>
Date: Thu, 22 Oct 2009 18:32:27 +0300
To: NGC4LIB_at_LISTSERV.ND.EDU
> -----Original Message-----
> From: Next generation catalogs for libraries 
> [mailto:NGC4LIB_at_LISTSERV.ND.EDU] On Behalf Of Karen Coyle
> Sent: 22 octombrie 2009 17:55
> 
> Dan Matei wrote:
> > Work:
> >
> > 	Title: 	The Conquest of Bread (en)
> > 	Title: 	Conquête du pain (fr)
> > 	Title: 	??? (ru)
> > 	Date:		1892 
> >   
> 
> Dan, this seems nice, but it isn't what is defined in FRBR or 
> RDA. The Work entity only has the "work title," which is the 
> original title. So you won't have these other titles in the 
> Work entity. If there is a Work with links to Expressions, 

No... The Work is language-independent (language comes-up only at the expression level). So, the
"work title" is "conventional" (like the uniform title). Whow knows the title Homer gave to Iliad ?

<citation source="FRBR report">
The title of the work is the word, phrase, or group of characters naming the work.
There may be one or more titles associated with a work. If the work has appeared
under varying titles (differing in form, language, etc.), a bibliographic agency
normally selects one of those titles as the basis of a “uniform title” for purposes of
consistency in naming and referencing the work. Other titles under which the work
has appeared may be treated as variant titles for the work, or in some cases as a
parallel uniform title. The title of a work that forms part of a larger work may
consist solely of a number or other generic designation that is dependent on the
title of the larger work.
</citation>

> then you could traverse that link and find the Expression 
> titles, which is what you have here. However, what I hear 
> from folks who do cataloging is that you are sitting there 
> with the manifestation in hand. If you aren't working in a 
> large database, or if you are the first one to catalog the 
> item (those two situations are functionally equivalent), all 
> the information you have is on that book you are holding. No 
> info on expressions, no info on Work.

The record for the manifestation is made (or could be made) nowadays directly by the publisher
(better or worse, but when the user can see the cover at Amazon, it is not so important that the
statement of resposibility is not transcribed canonically ?

I think the librarian has to do a bit of research in the autority files (which the semantic web
could upgade them to a "global knowledge base".

> 
> FRBR seems to work best when you already have a large 
> database of bibliographic data. You can then bring records 
> together based on the information in them as a collection.

Yes. The large database is called "the web".

> 
> When you say:
> >
> > There will be an expression entity linked from the Work,  
> via "is realised through" relationship:
> >   
> 
> That sounds good, but where is that happening? Take my 
> example: I have this Kropotkin book in English. I don't have 
> it in French and I don't even know that French is the 
> original language. What are the steps I take to create my 
> catalog record? Am I working in a huge database, like OCLC? 
> What if I am working in my local catalog?

Do not work in your local catalog ! Unless you have an "ivory catalog" :-)

> 
> I like the idea of relationships between WEMI entities, but I 
> fear that oftentimes they require knowledge that the 
> cataloger does not have and other records that the cataloger 
> does not have access to. I also don't know what those links 

Well, we have to learn :-)

> will consist of. I think the FRBR folks had in mind a 
> relational database model (as Tom Delsey often says), in 
> which the DBMS has a "private" way of linking tables that 
> isn't valid outside of the DBMS. But that doesn't allow us to 
> share those links, since in each of our systems they are 
> different. We CAN share names and subjects because we each 
> have the same link to the authority record (either the 
> authoritative string or the authority record number, usually 
> the former today).

(I used to be a teckie, so, I say) forget about the database organisation. Think "linked data" ! Sir
Timothy (BL) say so :-)

> 
> This essentially is my doubt about WEMI -- I can't figure out 
> how to create the entities in a sharable way that doesn't 
> require there to be one and only one system holding the entities.

Each antity (with a persistent URI), somewere in the cloud !

Dan
Received on Thu Oct 22 2009 - 11:34:22 EDT