Thomas wrote:
Somewhere in this mix there is the notion of the "primary work" (a phrase found in RDA at 20.2.1.1.). Some of the RDA expression attributes and relationship elements settle around an idea that there are supplementary works being expressed as augmentations to a primary work. RDA 7.15 has illustrative content as illustrating the "primary content" of a resource. So maybe there is a soft way of handling augmented expressions, with elements like relationship designators (like "illustrator") that convey this complexity in indirect ways. If there's a primary work, there would be a primary expression, and a cluster of elements that fall just outside this boundary but still forming a uniquely identified expression entity.
For real life cataloging I think this is a completely acceptable solution.
I like a collection of short stories example. Say there's a collection of short stories by Alice Munro that she wrote and selected to be published together, with a collective title. Is this one expression? Or is it a set of expressions, one expression for each short story?
I still feel that there is no need for having aggregate expressions at all (note that in my model there is no such thing). It only leads to a lot of problems and doesn't seem to solve any.
Perhaps at a later time there is a "Best of ... " volume that has some of the short stories from the earlier publication (and let's say it's Alice Munro doing all the selecting or "aggregating" in that case as well). If the collective works are named entities, and required for different kinds of relationships, then it's impractical to downplay this reality in saying that that aggregates only exist at the manifestation level.
Absolutely. But if we want to bring out the fact that an independently existing aggregate work is part of a new aggregate work, this can be modeled on the work level alone. Have a look again at fig. 3 in my "Additional diagrams #2" paper. The difference between the two "Selected works" editions of Jane Austen containing the same novels is depicted here through the parts of the two aggregate works (which lead to different expressions of the individual works). It's not exactly the same case, but the modeling could be done quite similarly for two collections containing a different set of short stories. And as the model is recursive (although the diagram then gets rather complicated), it would also be possible to have an aggregate work (such as an earlier collection) as a part of another aggregate work. There would still be, I believe, no need for aggregate expressions and whole/part relationships on the expression level.
Also I think the notion of "aggregating work" is problematic-- it becomes an empty shell entity in this case.
Not only there, I believe. I already put a long list of things which don't apply to an aggregating work. Now I even wonder: Can an aggregating work have a title? This certainly could not be the title of e.g. the collection or monographic series itself. It would have to be the title of the _process_ of aggregating things. It's really annoying that the final report doesn't give at least one example where they show the attributes of an aggregating work, and not only its relationships.
Heidrun -- --------------------- Prof. Heidrun Wiesenmueller M.A. Stuttgart Media University Faculty of Information and Communication Wolframstrasse 32, 70191 Stuttgart, Germany www.hdm-stuttgart.de/bi