Discussion paper: RDA and WGA treatment of aggregates

9 August 2016

Discussion paper: RDA and WGA Treatment of Aggregates


Discussion paper: RDA and WGA Treatment of Aggregates. Appendices


Submitted by: Deborah Fritz, Chair, RSC Aggregates Working Group


Print Friendly, PDF & Email

Tagged , , . Bookmark the permalink.

33 Responses to Discussion paper: RDA and WGA treatment of aggregates

  1. Tina Shrader says:

    General comment: I understand that this paper specifically excludes continuing resources, but given that FRBR-LRM treats serials as a type of aggregate, I think that the modeling in this document should at some point (perhaps by a different group) be extended to continuing resources.

    • Kathy Glennan says:

      General comment: note that in most places where the main paper references “Appendix C”, “Appendix B” is meant.

      General comment: I’m not clear on about what is meant by giving various options in the Appendix. I think it’s about cataloger’s judgment rather than a single binary choice that RDA needs to make, but that’s not completely clear.

      • Tina Shrader says:

        I found all the options in the appendix to be useful in exploring different ways in which the same basic model could be applied to the same situation. I don’t think RDA necessarily needs to make binary choices about which option is preferred, but I do think that communities of practice will need to develop their own guidelines and preferences to facilitate consistency and data sharing.

        • Chair says:

          [Comment from Deborah Fritz, AWG Chair] The various options in the examples in Appendix B are intended to illustrate cataloger’s judgment. We will try to make this clearer.

  2. Tina Shrader says:

    Responding to the questions posed in the paper:
    Question 1: Yes. I understand the AWG’s desire to maintain the distinction between whole/part works and aggregation works, but I think that providing unambiguous, easily applied guidelines for how to determine whether something is a whole/part or an aggregation may not be possible. The definitions provided in the paper make sense, but I question whether they are clear and unambiguous enough to form appropriate instructions for catalogers.

    Question 2: I think this is absolutely essential.

    Question 3: I agree that there needs to be relationship terminology that describes the relationship between a distinct work and the aggregation work, and I like the ‘incorporated in/incorporates’ terminology.

    Question 4: I like the ‘incorporates/incorporated in terminology. Not sure about where to place it in J.3

    Question 5: I agree that the relationship is a useful shortcut in some circumstances. I dislike the reciprocal value ‘creator of content of’. I understand the need for the reciprocal, but that phrasing is so semantically awkward that it makes no sense.

    Question 6: I think I agree with this, but I’d like a clearer definition of what they mean when they say, ‘unless the whole concept of the aggregation work has changed’.

    • Chair says:

      Passing along comments from Deborah Fritz, AWG Chair:

      re: Question 5–It might be clearer if you picture it as:
      • Work has creator of content Person
      • Person is creator of content of Work

      re: Question 6 — We will work on the wording, perhaps: “the addition, deletion, or revision of an included expression in an aggregation Expression requires a new aggregation Expression. A change in the concept or idea behind the aggregation work (e.g., entirely new subject matter, a radical change in format (a feature film based on text), or a change in geographic coverage, requires a new Aggregation Work.

  3. Diane Napert says:

    Question 1 – clear instructions as Tina mentions
    Going forward an example of a collection of poems and a collection of musical works on a sound disc would be helpful

  4. Diane Napert says:

    3. I think, in theory, the “incorporated in /incorporates” could be helpful

    Agree with Tina on 5, hard to understand, at least in my humble opinion Perhaps with an example I could wrap my head around it better

  5. Kathy Glennan says:

    Question 4:

    This needs to be in a separate section in J.3; it doesn’t belong in any of the existing categories.

  6. Kathy Glennan says:

    Question 5:

    I really need to be convinced that this type of relationship is useful. The definition of “creator” is already really “creator of content”. Isn’t it the “creator of aggregated work” that is the missing definition?

    This was even more jarring in the Appendix to the paper, which suggested using “author of content” (see p. 6). The rationale for this distinction is to “allow the user to distinguish between her [Austen’s] works and when her works are included in a compilation.” What’s the use case for this? How does it meet the various FRBR (or FRBR-LRM) user tasks?

    • Chair says:

      [Comment from Deborah Fritz, AWG Chair] No, that is just the issue: when we describe the aggregated work, the compiler of the aggregated work is its Creator. If we choose to describe one of the distinct work/expressions in the aggregation, then the creator of that work is its Creator. But the creator of one (or all) of the distinct work/expressions is not the creator of the Aggregation Work, so if we want to connect a creator of one (or all) of the distinct work/expressions, to the Aggregation Work, we have to relate that creator to the Expression of the Aggregation Work (since a new expression of the Aggregation Work might not include that creator’s contribution), and so need a new Expression relationship for this situation.

      This was one of the more difficult (radical) issues that the AWG discussed; so we may not be explaining it very well, yet.

      [Deborah Fritz, AWG Chair] The AWG will have to discuss the user tasks further, but it seems to me that the distinction between an “author of” relationship and “author of content of” relationship would meet the “Select” user task, i.e., select between a link directly to a work by the author (Author of), and a link to that work in a compilation (Author of content of):

      Austen, Jane, 1775-1817
      Author of: Austen, Jane, 1775-1817. Emma
      Author of content of: Complete illustrated novels (Austen)
      Instead of:
      Austen, Jane, 1775-1817
      Author of: Austen, Jane, 1775-1817. Emma
      Author of: Austen, Jane, 1775-1817. Complete illustrated novels

  7. Kathy Glennan says:

    Question 6:

    I agree with the 1st bullet.

    I’m not convinced about the 2nd bullet. Aggregation Works are works. Separate instructions in RDA are only needed if/when they get a different treatment from other types of works.

  8. Kathy Glennan says:

    Definition of Aggregation Work (Appendix A):

    Why not just: “… and/or arrangement of one or more expressions” (instead of “… and/or arrangement of expressions of one or more other works”)?

    • Chair says:

      [Comment from Deborah Fritz, AWG Chair] The wording for this comes from the FRBRoo definition of an Aggregation Work. We can’t select and/or arrange “one expression” but we can select and/or arrange multiple expressions of one work or more than one work.

  9. Kathy Glennan says:

    Definition of Single Work (Appendix A):

    This definition needs revision. I initially made something of a connection here to FRBRoo’s Manifestation Singleton, due to the phrasing “realized by one and only one distinct expression”. After I reviewed the examples, that’s clearly not what is meant.

    Use of “distinct” is a problem here as well — with the introduction of this terminology in the table on p. 1 of the appendix, it would be reasonable to assume that “distinct work expression” is meant. However, I’m fairly sure that’s not the case either. In fact, I think the entire appendix should be reviewed/revised so that “distinct” will be used in an unambiguous sense.

    Here’s an example of the problem, as I see it, with the current definition:

    Austen’s Emma clearly exists in multiple expressions (including translations). How then can it be considered “a work that is realized by one and only one distinct expression, embodied in a particular manifestation”?

    (The meaning of “single work” is better conveyed on p. 3 “the manifestation contains only one distinct expression of one work”. I think that rewording this concept in terms of “work” caused my problem.)

    • Tim Kiser says:

      I agree that the definition is troublesome for the reasons Kathy describes.

    • Chair says:

      [Comment from Deborah Fritz, AWG Chair] A “Distinct Work” is a work whose expression is incorporated in an expression of an Aggregation work. A “Distinct Work” can be a “Single Work”, a “Whole-Part Work” or an “Aggregation Work”

      [re: the ‘Emma example above’] This particular manifestation embodies a work that is realized by one and only one distinct expression. But how do we turn that into a definition for a “Single Work”?

      The AWG will try to further clarify the definitions of Single Work and Distinct Work before the Nov RSC meeting

  10. Mary Huismann says:

    Comment from a member of the music community: I’m not certain this is all fitting in my brain at once, but I do at least agree with the premise that always providing access to the creators of the content within aggregates is important. Beyond that it’s difficult to say what kind of effect this will have on us with the exploration of music materials at such a rudimentary stage. In particular, what current contributor relationships they decide should be ported to creator of content, and the degree of flexibility allowed to account for the conceptual models of different target audiences, will have a major effect on usability for us.

    Slightly tangential, but after wading through pages and pages of flowcharts and acronyms the sensation in the back of my mind is growing that no matter how elegantly we manage to map out these sometimes enormously complex relationships, it ultimately doesn’t matter if 1) no system makes use of it and 2) no user understands or cares about it. “RDA is a content standard, not a display standard” is a lovely mantra, but at some point it begins to sound a bit like “and then a miracle occurs.”

    • Mary Huismann says:

      Comments from the music community: This whole discussion paper and the accompanying appendices are a lot to chew on, and I can’t pretend to have processed it all. One thing did jump out at me. The AWG writes on page 9 of the discussion paper that “if a manifestation embodies multiple distinct expressions, then either an Aggregation Work or a whole-part Work and [n] distinct Works are present.” This means that a manifestation embodying parallel expressions of a single work is considered an Aggregation Work (as only one distinct work would be present). You can see this in Example 13 in Appendix B.

      I wish the discussion paper had mentioned parallel expressions, which are quite common in music – it is focused entirely on compilations or augmentations, and parallel expressions only arise in the appendices. The discussion paper draws a lot from the IFLA WG on Aggregates, which treats parallel expressions only briefly and implies that they are largely confined to manuals and government documents. FRBR-LRM does define, as a distinct entity, the “Aggregate of Parallel Expression.” I find this distinction useful, as I don’t think a manifestation embodying an opera in two languages should necessarily be treated the same way as a manifestation embodying a collection of multiple operas. Not all aggregates are created equal. The entity of Aggregation Work and Aggregation Work Expression seems more important for aggregate compilations than it does for aggregate parallel expressions.

      With regard to the last point (above), I agree that the complexity of some of these examples can get quite unwieldy, even for a FRBR nerd like me. I think, for example, that option 2 for modeling Appendix B example 13 is vastly preferable to option 1, and probably more in line with how patrons (and librarians) would mentally conceive of these relationships.

      • Mary Huismann says:

        Comments from a member of the music community: In the course of our (AWG) discussions, I brought up music issues. There will eventually be more music examples but we thought the models presented (there are others) were enough to digest for now. I am currently looking for some knotty music examples to model, so if you come across something, please let me know.

        Here is my understanding of how we might model scores and sound recordings:

        If a manifestation containing a score has the aggregation title “Symphony no. 5” (I did ask the group if they would be considering this at all if every other book was titled generically: Science text book, Mystery novel, Biography, Fantasy fiction, etc.) and includes a “significant” scholarly introduction, originally created in German, translated into French and English and “significant” critical commentary, same language scenario, the aggregated works/expressions could be: 1: “Symphony no. 5”; 2. Introduction (German); 3. Introduction (French translation); 4. Introduction (English translation); 5.-7. Critical commentary—same pattern. The creator of this kind of aggregation work could be an editor, a publisher, a foundation dedicated to the works the composer, etc. To distinguish the title from all the other “Symphony no. 5” aggregates created by a publisher the composer could be added: “Symphony no. 5 (Composer)” and so on for different manifestations that are viewed as aggregates by the same and other publishers. If the score had outstanding illustrations these could be aggregated, as could facsimiles, commentaries, indexes, work lists… anything one might consider significant.

        Now imagine a boxed set of 5 operas, with an aggregating title on the container, and for each opera you have a libretto in the original language with 2 translations and “significant” program notes in the same 3 languages. There could be 35 aggregated works/expressions of various sorts. Also keep in mind that right now we kind of fudge performances by cataloging them as musical works, but in the future we will probably account for them very differently and more aggregate interconnections may apply (have a look at FRBRoo F20-F26 and their links).

        • Mary Huismann says:

          Passing along another comment from the music community: A couple of comments from someone who barely understands the document:

          1. Figure 3 on page 7 of the discussion paper says “an analytical description of one of the distinct Works…”. But it looks like 2 works to me. Am I misunderstanding or is there an error?

          2. I found it disconcerting that definitions in Appendix A for “Whole-part Work” and “Whole-part Work Expression” are TBA, but I don’t have enough experience with these types of papers to know how common or accepted this is. How is a reader expected to understand the use of a term if the term is not defined?

      • Chair says:

        [Deborah Fritz, AWG Chair] The AWG noted that both the WGA report and LRM made a distinction between: Aggregate Collections of Expressions, Aggregates Resulting from Augmentation, and Aggregates of Parallel Expressions; but in the actual “Modelling Aggregates” section we still end up with only the choice of describing, separately, the “discrete” expressions embodied in the manifestation, and, if we choose, also the expression of the aggregating work embodied in that manifestation.

        Three options are available for each example, sometimes we chose to show only what we felt was the most obvious choice, other times we chose to show that different catalogers might make different choices. This was partly to raise the issue of how different descriptions of the same resources are to be reconciled.

    • Chair says:

      [Deborah Fritz, AWG Chair] Yes, we will have to examine all of the current contributor relationships to decide which of them should change to creator of content relationships and which remain contributor relationships

  11. Tina Shrader says:

    Passing along comments from CC:DA’s former chair, Dominique Bourassa:

    Questions 3 and 4:
    RDA already has relationship designators for “contained in (expression)” and “container of (expression).” What is the difference between these designators, if any, and “incorporated in” and “incorporates” relationships? Do we really need the new relationship designators proposed? If we do, we will need instructions that clearly show the differences between these two sets of designators.
    Question 5:
    This is an intesting suggestion. Example 11, option 1 in Appendix A shows the benefit of mapping such relationships when a choice is made of cataloging only the aggreation work and not the individual works. In RDA, there are currently some similar terms that can be found under relationships for works and expressions: “choreographer” and “choreographer (expression)”, “composer” and “composer (expression)”, “interviewee” and “interviewee (expression),” “photographer” and “photographer (expression),” etc. If we wanted to follow that pattern, the relationship would have to be “author (expression).” The problem is that these relationship designators usually refer to “supplementary content,” not “main content.” The author of content proposed in example 11, option 1 are part of the main content. Therefore, I am not sure what the proper term should be.

    Question 6:
    First bullet: yes
    Second bullet: I agree with Kathy that I don’t think it is necessary.

    Appendix A
    It is interesting to see all the different ways these resources have been mapped. It is a lot to digest. I find some of the mapping overly complicated. It would also be useful to see examples that use the different relationship designators for related expression currently in RDA J3 such as “augmentation of (expression),” “addenda to (expression),” “appendix to (expression), etc.” This might solve some mapping problems and simplify others.

    Example 4, option 2:
    In this example, Austen is not considered the agent agregating the work, therefore she is not the creator of the entire work, but becomes the the author of the content? This is against what RDA says to do. This means that if we follow that route, we would no longer record entire works of authors as:
    Authorized access point for the person. Works.
    This is a significant change in practice.

    Example 5, option 3:
    Here Blythe, is the editor and writer introduction of the AWE. In practice, I have nothing agains this. However, the definition of “writer of introduction” is “A person, family, or corporate body contributing to an expression of a work by providing an introduction to the original work ” (and not to a new “aggregation work.” Therefore, technically, we can only use this designator to show the relationship between Blythe and the distinct work expression for Emma, not to the aggregation work expression.

    Example 11:
    This is interesting. Considering Taylor as the agent of the agregating work seems to make sense (It is in line with what is typically done in bibliograhies). Someone has to have conceived this work and put it together. Taylor seems the logical agent. However, I am not sure the relationship designators “compiler” is meant to represent editors of compilation. Another problem is that if we accept Taylor as the compiler of the aggregation work, why not then consider in example 12, Regan as the agent responsible for the aggregation work Humhrey Clinker by Smollett + introduction by Lewis? If this is the route the RSC wants to take, we would need clear instructions explaining when an editor should be considered the agent responsible for the aggregation work (compiler) and when he/she should be considered a contributor to the expression. Finally, this goes against RDA that says the “if the work is a compilation of works by different persons, families, or corporate bodies, construct the authorized access point representing the work by using the preferred title for the compilation.”

    • Chair says:

      [Comments from Deborah Fritz, AWG Chair]

      [re: questions 3-4] The AWG thinks that the “contained in …” are for Whole-Part relationships. We agree that it is imperative to come up with definitions that will make the distinction between “whole-part” and “aggregation” very clear.

      [re: question 5] When we examine all of the current contributor relationships to decide which of them should change to creator of content relationships and which remain contributor relationships, the AWG will also make recommendations for changes to definitions. It is possible that we might suggest dropping the distinction between primary and supplementary content.

      [re: example 11] The editor relationship is separate from the compiler relationship. Additional evidence (in the preface and other places in the resource) suggested that Taylor was the compiler of the material; but it is quite possible that another cataloger would not agree with this; and I would not consider Regan the compiler for the Humphry Clinker example. might also have to be amended. The AWG is not ready to suggest changes to instructions yet, but we are pleased to see that you are noting the real implications of all of the aggregates models, including our own.

  12. Tina Shrader says:

    Passing along additional comments from colleagues at NLM:

    In the AWG version of modeling aggregates, if it is felt that “an aggregation work is not sufficiently significant to warrant bibliographic identification or description,” then later manifestations/expressions do warrant bibliographic identification, would we be obliged to go back and modify the original manifestation to add this information? What are display/retrieval impacts if different libraries make different choices on levels of description?

    I find the WGA Report Appendix B model more comprehensible that the main report model. Can someone please define what an “accidental whole-part” relationship that is mentioned in the FRBRoo model (at the very bottom of p. 3) might be?

  13. Tim Kiser says:

    Q1 & Q2: Yes.

  14. Tim Kiser says:

    Q3 & Q4: The concept is useful for describing aggregates; I would like further consideration of whether an existing relationship designator could be used for the purpose, per Dominique.

    If “contained in (expression)” and “container of (expression)” must be reserved for whole-part expression relationships, and a different set of terms must be used for aggregates, then I think “aggregated in/aggregation of” is more distinct from “contained in (expression)/container of (expression)” than “incorporated in/incorporates” (but only *slightly* more distinct, and not satisfyingly so).

    If the scope of J.3.4 cannot be expanded to include aggregates without muddying the distinction between whole-part and aggregate, then it appears that a separate section is necessary in J.3.

  15. kelleym says:

    1 Yes. The inferences that should be drawn in these situations are different so they should be distinctly labeled.

    2 Yes. If the distinction is kept, easily-understand definitions and heuristics are necessary.

    3 I have mixed feelings about this. I see the convenience factor, but I wonder if there are problems for inferencing similar to what happens when whole-part relationships are improperly applied. I find this hard to articulate, but I have put up a few diagrams at http://pages.uoregon.edu/kelleym/frbragg/aggregationExpressions.pdf that I hope convey what I’m thinking.

    The concern I have is that I think some of the conclusions that you want a machine to draw about the incorporated in relationship are asymmetrical. I don’t know if this is a problem for the modeling or not.

    1. If a user is looking for manifestations of expression X and X has a translation Y, you don’t want to offer the user manifestations of Y and vice versa.

    2. If a user is looking for manifestations of expression X and X has a whole-part relationship with Y, then it makes sense to offer up the manifestations of Y as a possible answer or partial answer to the user’s request (in my example, I think that should really be a W-W relationship, but I couldn’t think of an E-E example and this still makes the relevant point)

    3. If a user is looking for manifestations of a particular aggregating/aggregation expression X and the relationship is made at the E-M level, you get the correct inferences. It works whichever expression (or combination of expressions), the user is seeking.

    4. In the aggregation expression version, if the user starts from the distinct work expression, you do want to offer the user the manifestations related to the aggregation expression. However, if the user starts form the aggregation expression, you don’t want to offer all the manifestations related to the direct work expressions. Use cases for this might be someone seeking the Norton edition of Pride and Prejudice or someone seeking all the Norton editions edited by A.

    In the diagrams, the red lines represent cases where you would want to offer users manifestations attached to the related expression as well as the expression specifically sought by the user.

    4 I can’t think of a better label than “incorporated in / Incorporates” off the top of my head. I don’t think it fits in any of the existing categories in J.3 and should have its own section (like whole-part)

    5 If resources are going to be modeled this way, a RD is needed. For the Austen example, it seems to me that it would serve users better if there were a single whole work for the complete novels that could be re-used for every aggregating work. This would connect the aggregating work to the individual novels/parts without putting an undue burden on catalogers. This might be more problematic for living authors, though. The RD would still be useful for things like poetry collections or collected works where the individual contents are very numerous or infrequently collected.

    6 I have not thought through what all the implications are for this. Would translations of aggregation expressions also be handled this way? Are there consequences for inferences, such as what I pointed out in 3 above, that might be problematic in some way. Examples would be helpful here. Would all the different illustrated versions of a novel be aggregation expressions of the same aggregation work?

Leave a Reply