W3C home > Mailing lists > Public > public-xg-lld@w3.org > September 2010

Re: Non- and Partial-FRBR Metadata

From: Asaf Bartov <asaf.bartov@gmail.com>
Date: Thu, 30 Sep 2010 21:09:19 +0200
Message-ID: <AANLkTi=POSRKJ_mg-KRz75nE1gF3yYeU__LsnWCzrWiq@mail.gmail.com>
To: public-xg-lld@w3.org
On Tue, Sep 28, 2010 at 10:02 PM, Karen Coyle <kcoyle@kcoyle.net> wrote:

What I want is 1) to be able to say: This is bibliographic data, and 2) to
> link that bibliographic data to a Work entity even though it is *not* either
> an Expression nor a Manifestation. Ideally I would also like to make use of
> other relationships: like translationOf, versionOf. But those would be
> relationships between instances of karensBibData, not between karensBibData
> and the Work, in most cases. This would allow us to do things like link MARC
> records to wikipedia entries for a Work, as an example, and perhaps to even
> create interesting relationships between standard citations and library data
> without having to fully FRBR-ize everything. (This is sounding better and
> better to me.)

The FRBR separation of entities is precisely the conceptualization required
to facilitate expression of these "interesting relationships".  I wonder:
why do you say "either an Expression [or] a Manifestation" when you could
create both?  I.e. create two entities to represent your current Bibdata.
 Regardless of whether they are expressed as two entities in any user-facing
interface, the technical separation into E and M would allow you the
relationships you want.  Or am I missing something?

> yes, LCCN and OCLC number are identifiers for the record, or at least I've
> always considered them so. They are good clues to finding other instances of
> the same published "thing" but they represent the library record. The ISBN
> represents a publisher's product, and that product contains the whole WEM
> (but not Item, since ISBNs are not item-specific). Anything that is
> published is by definition a Manifestation of an Expression of a Work.

I think that's not quite accurate:
An ISBN corresponds (disregarding the problem of dupes) to a single
Manifestation, *not* necessarily to a W->E->M.  Remember the case of an
anthology, or an issue of a journal -- a single Manifestation embodying
multiple expressions, each of which realizes a single work.  So an ISBN
could really correspond to the follow set of FRBR entities
Thus your last sentence should probably read: "Anything that is published is
by definition a Manifestation of one or more Expressions, each of which
corresponding to one Work."

I am keen to hear if anyone disagrees with that.

It is unfortunate that in RDA, the publisher identifiers are defined as
> "identifier for the manifestation" -- which I think will be confusing.  (cf.
> Chapter 2 of RDA) This is where I just have a brain meltdown around FRBR,
> because in essence you cannot speak about, much less hold in your hand, any
> of the Group 1 entities by itself except at a very abstract level. Even the
> item, although it is concrete, cannot be separated from its co-entities.

While it's true that no single Group 1 entity would ever be something you
can "hold in your hand" in terms of library holdings, I do think they're not
too abstract to be interesting to consumers (albeit sophisticated, or
well-trained consumers).

Consider the example I mentioned in my introduction e-mail -- a 19th century
Hebrew translation of /Romeo and Juliet/ by Salkinson -- one of the earliest
translations of Shakespeare into Hebrew: clearly an Expression of the Work
"Romeo and Juliet".  There have been any number of doctoral dissertations
and articles written _about_ that translation (NB: not about /Romeo and
Juliet/), and so asking a system "What do you know about Salkinson's
translation of Romeo and Juliet?" could yield not only all the editions
(Manifestations) of the text the system knows about, but Works about the
Expression, and perhaps also Works about the Translator (but avoiding
flooding the user with Works about the Author, which are clearly mostly
irrelevant in such a specific query).

Or what if I ask my information system: "What can you offer by way
of related works to do with Romeo and Juliet?", and get a list of works
linked to the Work R&J, e.g. Tchaikovsky's ballet (and available renditions
in CD/VHS/DVD), /West Side Story/ (ditto), etc.

These seem to me useful options to have, as a client of information systems,
and I do not think it would be beyond the ability of library patrons and
consumers to get used to formulating such queries _when appropriate_ -- no
doubt full-text searching and brute keyword searching would still be primary
vectors of information retrieval much of the time.

Asaf Bartov <asaf.bartov@gmail.com>
Received on Thursday, 30 September 2010 19:10:13 UTC

This archive was generated by hypermail 2.4.0 : Friday, 17 January 2020 19:38:36 UTC