Re: Tim Berners-Lee on the Semantic Web

-----Original Message-----
From: Next generation catalogs for libraries [mailto:NGC4LIB_at_LISTSERV.ND.EDU] On Behalf Of Karen Coyle
Sent: Thursday, October 22, 2009 10:04 AM
To: NGC4LIB_at_LISTSERV.ND.EDU
Subject: Re: [NGC4LIB] Tim Berners-Lee on the Semantic Web
I'm not sure I understand everything Karen said, but I want to comment a bit anyway and focus on the Kropotkin example.

Karen said:

"Back to my Kropotkin example:

The Work has these elements:

    Author: Kropotkin, Petr Alekseevich, kni?az?, 1842-1921.
    Work Title: Conquête du pain.
    Topic: Communism

There will be a person entity linked to the Work:

    Person: Kropotkin, Petr Alekseevich, kni?az?, 1842-1921

There will be a topic entity linked to the Work:

    Topic: Communism

These properties link the Work entity with the person and topic entities."

Yes, but there is also at least another element the work _must_ have: it must be (or have been) embodied in an item. 
That item is best understand as one item from a set of duplicates, a manifestation (or impression or edition, to use older terms.) So that item has the following properties (among others)

"My Manifestation for the book has:

    Title: The Conquest of Bread
    Statement of responsibility: by Peter Kropotkin
    Place of publication: New York
    Publisher:Vanguard Press
    Date of publication:1926 "

So there is a logical connection between the item/manifestation and the other FRBR entities (WEMI, Persons, Subjects). Without the item (the embodied work) none of the other entities have any connection to one another.

"There is no logical connection between the Manifestation and any other 
WEMI entity. And it doesn't stand alone. Yes, I could add some 
identifiers, but they aren't inherent in the data itself so they would 
have to be artificial and not easily sharable."

I don't see why your item/manifestation description can't stand alone. It has the familiar critical elements of a citation. Those have stood alone for centuries. An item/manifestation description implies work, expression, person and subject information, and thus we'd be sensible to regard it as incomplete. However, that incompleteness doesn't prevent it from being a sufficient description--an effective, usable citation or reference to the item/manifestation.

"My guess is that the creators of FRBR were thinking along the lines of 
MARC -- a single bibliographic record with authority records for 
controlled headings. WEMI represents a single thing with four parts, not 
four things. That may help catalogers develop the record, and it may 
make it possible to merge groups of records into an interesting 
structure within, say, a single database, but you can't unleash WEMI 
into open space as individual entities and have them connect to each 
other. You CAN do that with W and Person and Topic, because they are 
logically interlinked by properties."

I come back to what I tried to say clearly yesterday about starting with the item--practically and logically without the item none of the rest (WEM, author, subject) exists. And thus, those things, (WEM, author, subject) make most sense as ideas we have about the items. Manifestation is an idea we have about an item that we know or suspect is not one-of-a-kind. This can be inferred or demonstrated. We think, "There are others like this item in that each shares a particular production history; in short, they were made together. Work is a further abstraction from the evidence the item provides. It is convenient for us to imagine an entity called "work" that allows us to intellectually connect many items that have the same content. Expression, for me, is best understood as a qualification of the work concept. Translations are a good example of our conception of a continuity of content despite enormous difference in expression (e.g. Plato's Republic in an English translation con!
 tains no word that Plato could have known yet we agree that the translation is still Plato's republic.)   

"Now the question is: can we fix this, or is it inherent in the data we 
are working with? I have to hope for another sleepless night."

I don't think the problem before us is "can we fix this (FRBR's WEMI structure). I rather see us ask, given that FRBR's WEMI structure provides us with a relatively formal and rigorous description of textual materials, how can we use it to greatest immediate and longer term advantage? I think one answer so far has been to help reconceived the rules for cataloging. Another has been to help us create better catalogs--catalogs that are better at using relationships to collocate, distinguish and display materials in collections.  


Matthew Beacom