“An aggregate is defined as a manifestation embodying multiple distinct expressions. Three distinct types of aggregates exist:”
-
Aggregate Collections of Expressions
-
Aggregates Resulting from Augmentation
-
Aggregates of Parallel Expressions
p. 64-66
FRBR Open Comments 
Pingback: World-wide review of the FRBR-Library Reference Model, a consolidation of the FRBR, FRAD and FRSAD conceptual models | FRBR Open Comments
FRBR-LRM is highly contradictory in its treatment of aggregates.
First, let me recap the “Final report of the Working Group on Aggregates” from 2011:
Click to access AggregatesFinalReport.pdf
Remarkably, this group wasn’t able to agree on how to model aggregates. So, in the main part of the report, one way of modeling ist explained. I’m going to call this “Model A”. However, in appendix B (p. 18-20), an “alternative approach” is explained (which was adhered to by a significant part of the group, as somebody once told me). I’m going to call this “Model B”. The gist of Model A, as I understand it, is that there are no aggregate works. I repeat, to let it sink in: There are no aggregate works. They simply do not exist.
If we look at a collection of twelve articles, e.g. in a Festschrift, we do not have *one* aggregate work that is made up of *twelve* single works. Instead, what we have is an aggregate manifestation. This aggregate manifestation embodies not twelve, but thirteen works: the twelve articles plus another work which is called an *aggregating work* (note the difference between “aggregate” and “aggregating”). This so-called aggregating work represents the creative effort of the editor in planning and bringing together the twelve articles. It is important to bear in mind that this aggregating work is not made up of the twelve articles (as an aggregate work would be). It has been called “the glue, binding, or the mortar that transforms a set of individual expressions into an aggregation” (Final report, p. 5). I’m sorry to say that I’ve always found this approach to be completely counterintuitive (not to say misguided).
In model A, whole/part relationships can only exist for works which are not aggregates. In their proposed addition to FRBR, team A writes: “A work may consist of intellectually or artistically discrete components, such as a chapter of a report, a segment of a map, a table from a report, etc.” (p. 6). Another example would be a novel with twenty chapters. Each chapter is a component of the work. However, the novel is not an aggregate – nor is the report or the map.
Model B, on the other hand, accepts the existence of aggregate works. An aggregate entity is “the ‘whole’ in a ‘whole/part’ relationship with two or more components (parts).” (p. 19). They go on: “The whole and its parts are connected through a ‘contains/contained by’ relationship, which could also be expressed as ‘has part/is part of.’ So when you are considering an aggregate entity, you have 1) a whole (aggregate), 2) its parts (components), and 3) the whole/part relationship between them.” The following text is a bit long-winded, but it is made clear that aggregate works do indeed exist, and that it is up to catalogers to decide whether they want to work on the aggregate level or on the level of the individual works: “In other words, we may choose to recognize a whole entity as an integral unit (e.g., a work treated as one unit although it may consist of the collaborative work of several creators), or we may choose to recognize the whole entity and its component parts (e.g., an aggregate work/whole, such as a trilogy of stories, with the parts being the component works, that is, each of the individual stories in the trilogy) in a whole/part relationship.”
I believe this is exactly the same as is expressed in RDA 5.1.2: “The terms work and expression can refer to individual entities, aggregates, or components of these entities (e.g., the term work can refer to an individual work, an aggregate work, or a component of a work).”
To me, model B makes much more sense than model A (although I accept that there are some difficulties with this approach as well). The idea behind it has often been called a “work of works”, i.e. we have a work which is made up of other works.
Now let’s look at FRBR-LRM. The draft has a separate section on the modeling of aggregates (p. 64f.). Here, the authors repeat Model A. I haven’t checked in detail, but I think they are even using the same words as the final report of the Aggregates Working Group. This means that FRBR-LRM does not accept the existence of aggregate works which contain other works, right?
But now let’s have a look at the main body of the draft, the tables for entities, attributes and relationships. Here we find the relationship LRM-R20, which is a whole/part relationship between two works. “Examples include movements of concertos, poems within poetry cycles, multipart novels, triptychs” (p. 51). I think the concerto example is similar to a chapter in a novel, so this is no contradiction to model A. But a poetry cycle made up of individual poems (which are works in their own right), a multipart novel like Ursula K. Le Guin’s Earthsea trilogy (which is used as an example in LRM-R21) or a triptych consisting of three paintings are certainly perfect examples for aggregate works, which shouldn’t exist according to model B. Also, the Wall Street Journal (p. 14) is used as an example for a work (cf. also 5.6 Serials). On the other hand, that I haven’t found any trace of an *aggregating* work in the main body of the draft. So although the authors of FRBR-LRM claim to follow Model A, in fact they model aggregates according to Model B. I cannot help but think that this is highly contradictory.
Heidrun Wiesenmueller
LikeLike
My best interpretation is that the whole-part examples given in FRBR-LRM are meant to be cases where the whole and part are intrinsically related. The movements of a concerto, however they are published, are always part of that concerto and the concerto always has those movements as parts. This seems clear with the Le Guin example. With the others, the assumption would seem to be that concertos are not revised to consist of different movements nor are poem cycles redone to consist of a different set of poems (or if this is done, the new versions constitute new works rather than new expressions). I don’t know enough about either of these forms to say.
An aggregate work, on the other hand, would seem to be a collection of things brought together contingently. A chapter in an edited book or a story in a short story collection could be published as part of other collections. Some expressions or manifestations of the chapter might not be related to the edited book so if the edited book were treated as an aggregate work related to the chapter work this would lead to incorrect inferences (as we discussed on RDA-L http://lists.ala.org/sympa/arc/rda-l/2016-03/thrd2.html#00026). I know that you are cognizant of this issue based on your own proposals for implementing an aggregate work. However, in order to have correctly-functioning aggregate works, I think you have to have multiple types or layers of works (as FRBRoo does with its complex work and individual work or as you laid out in your proposal). Since the plain vanilla FRBR/FRBR-LRM model does not include this, I think FRBR-LRM is forced to use model A. I don’t know that an aggregate work is necessary, but it could reduce some redundancy in recording relationships. However, I think the model would have to be expanded to explicitly incorporate multiple types of works if it is to support aggregate works.
LikeLike