Quoting "Brenndorfer, Thomas" <tbrenndor...@library.guelph.on.ca>:
Because there are many-to-many relationships that are horizontal as
well as the many-to-many primary relationship unique to expressions
and manifestations.
For the case of the aggregating expression, going over it again ...
An expression may be embodied in a manifestation. That expression
could be accompanied by other expressions embodied in the same
manifestation. That's a vertical many-to-one scenario--
expression(s) to manifestation. One doesn't need a special
whole/part relationship for this, because the primary relationship
(a vertical relationship) establishes this.
The resulting aggregating expression has whole-part relationships to
each constituent expression. That's a horizontal many-to-one scenario.
Following this logic, I doubt there would be a
manifestation/manifestation whole/part, except in perhaps a "bound
with" situation. Do you agree?
And it also seems that in your scenario, aggregates link whole/part
between expressions but not between works? Is there a reason why they
would not link at the work?
I did a very ugly diagram of this...
http://kcoyle.net/temp/frbragg.pdf
If it's too ugly I can try a do-over.
kc
That aggregating expression also has vertical relationship to the
manifestation. It could have another vertical relationship to
another manifestation (an e-resource version of a print publication
for example). The two types of relationships (horizontal and
vertical) are like check-off lists for all the possible
bibliographic relationships.
Going further ...
That original single expression may also have been published in a
>>different set<< of expressions in another manifestation (creating
a different aggregating expression). That original expression may
have other structural or content relationships to yet other
expressions (different horizontal relationships, such as 'adaptation
of' or 'revised as' or 'translation of'). More many-to-many
relationships are being fulfilled.
That original expression could also be published in different
manifestations (different vertical relationships -- now we're in
that vertical many-to-many territory). But that original expression
will only ever "realize" one and only work (and inherit its
attributes). So horizontally and vertically, that original
expression can have multiple relationships, except upwards to the
one work that it realizes.
From a display point of view, there are a lot of variables to consider.
In browse lists currently, analytical title entries are confusing.
In author browses, name-title headings juxtapose with the titles
found under the author's name. Title browses can be a bit better,
but the nature of the relationship of one title to another is not
clear until each record is examined in more detail. Keyword searches
will bring up brief record displays where the analytical titles in
the Content notes or added entries are obscured.
The LibraryThing approach is what I favor -- have a web page for
each entity, and populate that web page with all the attributes and
relationships associated with that entity. If it's the work, show
the available expressions (and manifestations), but also show the
horizontal structural and content relationships, so one can navigate
both vertically and horizontally. There is some redundancy in
navigating horizontally to the aggregating work and then down to the
same manifestation, but then one would be able to see all the other
associated works in the aggregating work, and explore outwards from
them.
Smarter systems may even highlight these kinds of whole-part and
vertical relationships, and perhaps go beyond LibraryThing by
tackling manifestation and item records. In my system, availability
information at the item level already influences screen display--
the user is directed to the available copies first. Our NoveList
widget alerts users to similar titles based upon presence in the
catalog (this is at the manifestation level). If all the FRBR
relationships were present, the system could alert the presence of
associated works at the highest level and direct users to particular
combinations of relationships, with appropriate deduplication. (Some
of the FRBR-lite utilities like LibraryThing for Libraries already
do some of this).
Thomas Brenndorfer
Guelph Public Library
--
Karen Coyle
kco...@kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet