RDA Entity Boundaries

Gordon Dunsire
RSC Technical Team Liaison Officer
12 June 2020

RDA Entity Boundaries

The paper proposes the revision and relocation of existing Toolkit content, and poses questions on
the coverage and granularity of Toolkit content on this topic.

Print Friendly, PDF & Email

Tagged , . Bookmark the permalink.

8 Responses to RDA Entity Boundaries

  1. Ryan Tamares says:

    Comment from AALL:

    Re: conformance. The reason people are asking about this is about RDA, just like the current, aka, almost old, RDA toolkit, is the level of abstraction. A conformance protocol would, seemingly, code in some more, to use programming, produce some statement on whether your metadata is ‘well-formed’ against some kind of schema. This makes sense kind of, But we all know there are exceptions to every rule. One opinion is that the guidance toward conformance should continue as an ongoing effort within the toolkit, but not as a coded situation outside of, say, application profiles tweaked by an institution. Our metadata does need some room to breathe.
    In this context, an AALL respondent states: no, RSC should not offer conformance certification in addition to the passive guidance included in the Toolkit. The same respondent also stated, yes, the Toolkit include content for assessing conformance for specific application communities. And for the question: Should the RSC offer additional technical guidance and support for conformance, interoperability, and mappings between RDA metadata and non-RDA applications?, the respondent states “a YES on question 2, with a note that I will need to think about 3 more”.

    Re: entity boundaries: I think I support what I see, we KNOW from experience that issues duplication come about all the time for all kinds of reasons, and there is no reason to think that can be solved – especially in light of rare or special materials held locally. so, I appreciate they are thinking about clarity on this now as this move to the new toolkit carries on.

  2. Thomas Dousa says:

    A common locution in the Condition statements for entity boundary conditions is “There is a significant difference in the value of an [Entity Type X]” and it is the presence of the “significant difference” that typically serves as the justification for recording a new instance of the entity type in question. What, however, constitutes a “significant difference”? In the introductory section on “Entity boundaries” on p. 5 of the discussion document, the reader is presented with a negative definition of what a significant difference is (i.e., “Differences in values of an element are not treated as significant when they are the results of applying different recording methods, options within a recording method, or different levels of granularity.”) However, no _positive_ definition of “significant difference” is given so that the reader must divine the definition of “significant difference” from that of “not significant” difference. (in other words, he or she must infer the definition as a negation of what a “significant difference” is not.) One hopes that all readers read the instructions carefully enough to make this mental switch but one also fears that many may not. Given that most of the guidance directs the cataloger to ascertain whether there is a “significant difference” in value, I think that it would be most helpful if a positive definition of “significant difference” were given in the introduction: this would certainly ease the cognitive burden of the reader.

    More generally, I admire the ingenuity with which the authors have pitched at a very high level of abstraction the traditional question of when new records (or rather, new metadata description sets) should be made for an item. My one concern is that the discussion is pitched at such a high level of abstraction that many working catalogers will find the instructions hard to follow. However, I tend to be a pessimist in hermeneutical matters and would be delighted to be proved wrong.

  3. Thomas Dousa says:

    There is another point where the introductory discussion of entity boundaries on page 5 would admit of greater clarity. In the element instructions for the discernment of boundary conditions for various entities, a distinction is made between “relative boundaries” and “absolute boundaries”. These terms, however, are not defined _expressis verbis_ in the introductory general guidance and so their meaning may be opaque for those users of the new Toolkit not steeped in the technical terminology of RDA 2.0. Presumably, “absolute boundaries” pertain to “criteria are required for conformance with wellformed RDA metadata and are determined by the semantic integrity of the type of entity”, while “relative boundaries” have to do with “criteria are specified within an application and are determined according to “bibliographic and cultural conventions””. If this is, indeed, the correct interpretation, then it would be good to associate the terms “absolute boundaries” and “relative boundaries” explicitly with the corresponding kinds of criteria in the introduction. If, however, the interpretation is wrong, it is still important to give an explicit account of what the difference between “absolute boundaries” and “relative boundaries” are (i.e., why some boundaries are “absolute” and others are “relative”).

    In general, my sense is that the abstractness of the vocabulary is likely going to be the greatest obstacle to uptake of the new RDA by many members of the cataloging community: for this reason, it is imperative that that all technical terms used in specific element instructions be clearly defined in the general guidance sections.

  4. Nancy Mitchell Poehlmann says:

    I will echo Ryan’s caution on entity boundaries, especially for rare books and special collections. It is not possible to remove all unnecessary duplication without removing some necessary refining elements.

  5. Keith Knop says:

    Responses from MLA:

    Q1: There was a general feeling that associating boundary criteria with specific elements would be good. It was also pointed out that even though the boundary of a manifestation may not technically need to be associated with specific elements, an element-by-element list could still be helpful in much the same way as OCLC’s field-by-field guidelines for when to input a new record.

    Q2: This is difficult to say without seeing a real application profile in a live environment, but option (b) seems most preferable.

    Boundary of a person could perhaps use some refinement. A respondent offered the real world example of two flute players named Leslie Newman, time/place of birth unknown but active at the same time. As written these guidelines seem to suggest these should be treated as the same person even though that is surely not the intended result. One can infer that, as these can be determined to be two different people for reasons NOT considered boundary criteria, they cannot share the exact same time and place of birth, but that seems like a backwards way of arriving at a simple conclusion.

    The terms “absolute boundary” and “relative boundary” are not defined anywhere in the paper, nor do those terms seem to have prior evidence in the Santiago paper, LRM, or the extant beta Toolkit text. If this is intended to serve as a guidance document readers should have all relevant terminology clearly defined.

    Some other tangential issues came up in response to this paper. Transformation by genre and transformation by style are both possible boundaries for works, but RDA does not define either “genre” or “style,” both of which can have widely varying meanings depending on context. If the intent is for those terms to be defined by communities of practice it would be nice to note so in their respective element pages or in the glossary.

  6. Amanda Ros says:

    Comments from Mark Ehlert:

    1. Appendix 2 is missing “Boundary of a family.” Is this an error?

    2. I agree with Keith Knop’s observations about capturing the meaning of “style”, less so with “genre.” As a monograph cataloger, I also add “editorial policy.” Glossary definitions of “transformation by …” and “transformation by … of” elements are quite circular. Refining these definitions would introduce clarity. For instance, the “transformation by style” element (https://beta.rdatoolkit.org/Content/Index?externalId=en-US_ala-8b78accc-ad68-38b6-b759-887f370af8ce) lists the following narrower elements at the bottom of its page: freely translated as work, imitated as work, paraphrased as work, remade as work. Ah, so that’s what “style” is aiming at.

  7. Kathryn Lybarger says:

    Should “Boundary of a Person” also include a condition for the persons’ names being different? (Are these conditions listed the only ones we can use to distinguish between persons?)

    I’m finding this pretty hard to read. Are the directions in the appendix meant to be read / applied by catalogers, or is this general guidance? What do you actually do if a difference is “determined by bibliographic and cultural conventions”? (Is that for application profiles to define?)

    I think somewhere we do need element-by-element lists along the lines of “Differences Between, Changes Within” or OCLC’s “When to Input a New Record”.

  8. Thomas Dousa says:

    This comment pertains to the instructions for “Boundary of an item” on p. 12 of the draft. According to these instructions, a new metadata description set should be made for an item if there is a “significant difference” in the value of a modifying agent (or subtype) or in the “accompanied by” element (or subtype). This raises two questions:

    1. Suppose that a library has two exemplars of a given manifestation that are unmodified in any way and do not have any accompanying materials. One of these exemplars is located in the general collection of the library while another is housed in the library’s special collection. In this case, neither of the stated conditions for creating a new item record is met and so there would be no trigger for creating a new record for the item. Indeed, one could say that the two items are identical in all respects save location (i.e., each occupies a different segment of space-time): that is to say, they are qualitatively identical, but numerically identical. Why, then, is not location (or, better, a relation to Place) not also among the conditions for creating a new record for an item?

    I suspect that the reason for not including location is that this would lead to a hypertrophy of metadata description sets, since, in principle, every exemplar of a given manifestation would then require its own record, since each item occupies its own unique bit of spacetime. This, however, raises another significant question about what is meant by “recording a new instance of item.” Any given metadata description set about a given information resource is going to include statements about a Work, about an Expression, about a Manifestation, and, perhaps, about an item: let’s call these “Work-level”, “Expression-level”, “Manifestation-level”, and “Item-level” statements. Does the expression “recording a new instance of item” mean “creating Item-level metadata statements about the information resource”? Or does it mean something else? The way in which the instructions have been formulated is far from clear.

    2. Suppose that I have two exemplars of a book signed by a single author, one exemplar of which includes an inscription by the author consisting of the author’s signature alone, while the other includes an inscription by the author consisting of a long dedication to a famous personage by the author. Let us also suppose that, apart from this variant in presence or lack of dedication, there is not other modification of the book in question. According to the current instructions, because there is no “significant difference” in the modifying agent (i.e., the author) nor in the “accompanied by an item” element, there is no warrant for recording a new instance of the item. And yet, for many catalogers, the difference in the content of the modification of the two items might seem significant enough to justify the creation of a new record: after all, the addition of the dedication associates the second copy with another agent (i.e., the famous personage to whom the item is dedicated) in a way that the simple signature fo the first copy does not. One wonders, then, if the current set of conditions allowing for the recording of a new instance of item are not perhaps overly restrictive.

Leave a Reply