Thanks for this thoughtful reply, John. As you can tell, it's taken me
some time to contemplate the implications of all of this. I probably
have MORE questions now, but maybe they are more to the point.
Quoting John Attig <[email protected]>:
The designations are very explicit about the nature of what might be
called the "target" entity; even for the same relationships (e.g.,
"based on"), the designators are qualified to distinguish between a
related work and a related expression. However, the nature of what
might be called the "source" entity is NOT specified, so "based on
(work)" could be a work-to-work relationship or an expression-to-work
relationship. A "source" entity can be either a work, an expression, a
manifestation, or an item.
It's actually the opposite which makes more sense to me... from a
practical point of view: You have in hand the "source" -- most likely
a manifestation/item. It seems that you would know more about that
source than you would about the "target." So I could easily have in
hand a translation without knowing WHICH expression it is a
translation of. Or I could have an adaptation and not know if it is
adapted from the Work or an Expression (I believe adaptation can link
to either). I might need to say that I know that my manifestation is
an adaptation, but I don't know what the adaptation was based on.
It also seems to me that there are some logical entity-entity
relationships and some illogical ones. One seems to have two choices:
control the "pairs", or leave the pairings open allow users to combine
them based on their own judgment. The controlling of one entity in the
relationship but not both... I'm still trying to find the reasoning in
that. Or more importantly, the practical purpose.
In a scenario 1 implementation, the nature of the "source" entity is
implicit in the type of record used to record the relationship; if the
"source" entity is described in a Work record, it is a work, etc.
Well, this rather contradicts RDA's goal of being implementation
neutral -- it assumes that there is a Work "record", and Expression
"record", etc. That is pretty specific as to design. I can easily
imagine a data structure in which the WEMI information is present but
not in separate records. As a matter of fact, many if not most of the
RDA data elements are specific to a particular FRBR entity. Thus
"title of the Work" itself has "Work-ness" inherent in it. In semantic
web terms, data elements are often associated with particular classes,
and in the registered RDA elements the FRBR entities are defined as
classes. Each RDA element "belongs" to one of those classes. Perhaps
the better way to express this is to say that the nature of the
"source" is inherent in the class membership of the source. That way
one isn't presuming that the entities are each expressed as a record.
Which doesn't eliminate my confusion about the binding of the pairs to
particular entities...
3) I don't find some relationships that seem to be key:
- Expression of
- Manifestation of
- Item of
The relationships between work, expression, manifestation and item for
a given resource are defined as Primary Relationships and are dealt
with in Chapter 17 of RDA. The relationships in question are treated
in RDA as elements; relationship designations are not provided for
elements; the element name itself designates the relationship.
By elements do you mean that they are on the element list? Because I
don't find them there. Or by elements are you referring to the
presumed "record"?
These are key relationships, and I don't see how they can be treated
as anything BUT relationships. How would you connect a Manifestation
to one or more Expressions, if not using a relationship?
- Translation of (Expression as translation of Work)
Translation of/Translated as is covered in Appendix J under J.3.2.
Note, however, that Work is an abstract entity that has no linguistic
content and therefore no language; the translation relationship is
always between two expressions, e.g., an English translation of the
original German expression of the work Goethe's Faust. Our current
practice obscures this fact by using the same access point to name both
the work and its expression in its original language; technically, we
should (and hopefully will) distinguish between Goethe. Faust (the
work) and Goethe. Faust. German (the original expression).
Ah, I was looking for this in the wrong place! This brings up for me
something I have pondered in the past: the "original expression"
question. There doesn't seem to be any concept of original expression,
yet that information may be needed for some relationships. Did JSC
consider having a "privileged" Expression that represents the original
text?
kc
--
Karen Coyle
[email protected] http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet
