I don’t think that the discussion is getting antagonistic, though I do appreciate the efforts to keep it that way. So I will attempt an answer to Kevin’s question as best I can.
In essence, I believe Kevin’s question is, When the PCC adopts the post-3R RDA Toolkit, can we just continue to call non-RDA entities creator? The answer is yes and no.
We could continue to use an entity whose definition includes real, fictional, legendary, and non-human entities, and elements which specifies a creator or contributor relationship to a work.
What we will not be able to do is use the RDA entity ‘agent’ and the _RDA_ elements ‘creator agent of work’, etc. for this mix of real humans and others.
When we make a statement such as (pardon my pseudocode), “rda:work rda:has_creator_agent_of_work rda:agent”, we are making a declaration to the world that the subject and object of the statement meet certain definitions. If we violate those definitions, we will cause problems for anyone trying to use our metadata. It would be like deliberately entering LCSH headings into a 650 with second indicator 2 for MESH. We would be telling the world that this is a MESH heading, but it does not conform to MESH rules. Anyone else trying to use that metadata, in particular to repurpose the metadata for other purposes, would have a problem with the incorrectly coded headings.
We do not have to use the element ‘rda:creator_agent_of_work’. We could use elements from other models, or define and maintain one ourselves, in much the same way that one can use multiple schemas in an XML document. In fact, we are doing that already with some pieces of metadata which do not fit into RDA. But when we adopt the post-3R RDA Toolkit, if we decide to use ‘rda:creator_agent_of_work’, I believe we really must use the RDA definition or else face significant problems interfacing with other schemas, making our metadata essentially useless in the linked data universe. It is not a question of writing ‘true rda’, as Kevin puts it. It is about compatibility, repurposing of data, and linking our data to the rest of the data universe. When we say a piece of metadata works a certain way, other people rightfully expect it to work that way.
Stephen’s question deserves an answer. The assertion was made that RDA, at least on the matter in question (non-human agents) was simply aligning itself with other international standards (“harmonizing” with other models, including museums). I remember that during this almost exact same discussion several years ago it was pointed out that many other international models do permit non-humans to be considered agents. In other words, at least at the time, the assertion that we had to change in order to align ourselves with what everybody else was doing rang more than a bit hollow. But I am not at all informed about the current state of affairs with other models. So Stephen’s question, “how do museums, or other non-library agencies, handle fictitious entities and non-human entities (including AI, animals, etc.), if at all?”, does deserve an answer.
Robert L. Maxwell
Ancient Languages and Special Collections Librarian
6728 Harold B. Lee Library
Brigham Young University
Provo, UT 84602
Pardon for the repeat, but I want to rephrase this more precisely:
So how do museums, or other non-library agencies handle fictitious entities and non-human entities (including AI, animals, etc.) , if at all? Maybe we could learn from their experiences?
Thanks for this clear explanation.
So how do museums, or other non-library agencies handle fictitious entities, if at all? Maybe we could learn from their experiences?
I’d like to step back for a minute here. My assumption is that we want to create metadata statements that follow a standard. To give a metadata statement validity outside of our local context, we need to identify the standard used. Different metadata statements in a description set can use different standards; we do this all the time (RDA + LCSH, etc.). Because of our training and the overall context, we have an implicit understanding of which elements in a MARC 21 Bibliographic record are associated with a particular standard; some of this identification is explicit, but some is not. We also recognize that it is important to accurately identify which standard is being used; it’s not useful to misidentify MeSH as LCSH, or Dublin Core as RDA.
The RDA element set has been developed based on the semantic web, with an assumption that libraries want to make their resources discoverable on the open web, and not exclusively through our carefully curated silos. RDA is an implementation of the IFLA Library Reference Model, which was informed by work that had already been done by CIDOC (The International Council of Museums Committee on Documentation) and IFLA in creating the object-oriented version of FRBR (FRBRoo). Work first started on harmonizing these library and museum conceptual models in 2000. The much-discussed change of “agent” no longer encompassing fictitious entities is an outgrowth of this harmonization.
As PCC proceeds in determining how to implement the post-3R RDA, we will make important decisions about elements to require and to ignore, and we may also wish to make additional refinements. I strongly recommend that we do not make these decisions on the basis of what we currently do with MARC 21 records, our legacy data, or our systems, and how we understand these today. For example, we will need to make a decision about the nature of our authority records: are they controlling the use of a given name, or are they collections of data about a person? This difference has a profound impact on how we treat pseudonyms (e.g., one “record” for Mark Twain, or separate “records” for Twain / Clemens / Snodgrass / etc.). As many of the institutions participating in the PCC Wikidata Pilot have discovered, when we expose our data on the open web, we are no longer fully in control of the content, use, or even the guidelines applied to a given piece of data.
It is certainly PCC’s choice about how much of RDA to implement. However, it will also be important for PCC to develop, document, and maintain any alternative standards that are used in the place of existing RDA elements. PCC will need to seriously consider the costs and benefits of such alternative approaches, not just for catalogers, but also in terms of how various types of descriptive metadata elements will play together on the open web.
On Mon, Mar 15, 2021 at 2:15 PM McDonald, Stephen <[log in to unmask]> wrote:
The report of the PCC Task Group for Coding non-RDA Entities in NARs does not say anything about recording relationships. It is exclusively focused on how to code these entities in the Name Authorities File. Recording relationships is part of the discussion on attributions, which was determined to be outside the scope of that task group. I don’t know of any PCC proposal yet for recording under LRM the relationship of a non-RDA entity attributed to be a creator or contributor. I don’t think those policies have been developed yet.
When I talked about "true" RDA, I didn't mean that every single element in a record would be an RDA element. What I meant was that there would be nothing in the record that would *violate* RDA. That is, a "true RDA" record may contain non-RDA elements, but all RDA elements would be used only in ways authorized by RDA. That is a very significant difference.
What PCC is proposing is an elaborate mechanism to make our records "true RDA" as I describe above, by using different relationship elements to relate WEMI to fictitious and non-human entities than those used to relate WEMI to RDA agents. This is going to require complex programming in LMS and discovery systems to make the different classes of relationships (RDA and non-RDA) work together seamlessly and transparently for library workers and library users. After all of the progress we have made over the years in getting our metadata to represent the resources in ways our users expect, we are going to be going backward and making the metadata more vague and/or difficult to search, navigate, interpret, and manipulate.
I would much rather that PCC choose instead to ignore RDA's restrictions by either allowing fictitious and non-human entities to be considered RDA agents, or allowing RDA relationship elements for agents to be used with non-RDA entities (such as fictitious and non-human entities).
Kevin M. Randall
Principal Serials Cataloger
Northwestern University Libraries
Proudly wearing the sensible shoes since 1978!