Notes onMoving Away from MARC-a-thon Discussion Questions May 4, 2016

Resource 1: single part print monograph

  • Question about recording multiple roles in the BF Editor. Answer: to enter the person once with multiple roles rather than enter the person twice.
  • Question/Discussion about subjects entered as pre-coordinated or separate elements. Currently the BF editor cannot create RDF data for pre-coordinated subject strings that do not have their own authority record in id.loc.gov. This would be the case for all free-floating subdivisions. The subject must be entered manually as a string and is treated as a literal. There was discussion of whether LCSH would be better served by breaking the subject string apart into separate elements to allow for better linking. Note: This is not a problem with MeSH as there are no free-floating subdivisions and every allowable combination has its own URI. All geographic and publication types are also entered separately and would be able to link to their respective URIs.
  • Question/Discussion about actual dates recorded as text strings versus recording ISO style dates. The benefit of recoding ISO style dates is that they could have individual URIs defined for them rather than literal text strings. There was discussion about implied publication dates in square brackets which are not useful in a linked data world and should be converted to actual dates. There was also discussion about copyright dates with the copyright and phonograph symbols which causes these dates to be read as literals. The issue of copyright dates is under discussion at the RDA Steering Committee (RSC).
  • Question about punctuation in the statement of responsibility element. Answer: it is still needed if there are multiple statements in one element.
  • Question about multiple places of publication. Answer: they can be entered in the same statement with punctuation in between, or in separate elements.
  • Question about how to handle initial articles. Answer: leave them in and a system can be programmed to ignore them if you need to create a browse-type list.
  • Question: Was difficult to comprehend the describing of the resource without using MARC language?Answer: Consensus was No.
  • Comment that some RDA elements are hard to understand and determine what information should go where. UC Davis performed a pilot experiment with some catalogers and will forward their results on this topic.
  • Question: Was knowledge of RDA entities was useful for the discussion?Answer: consensus was Yes, a better understanding of RDA and FRBR leads to a better understanding of the new tools which will make the transition away from MARC easier.

Submitted by Barbara Bushman

Resource 2: Online serial

Question 1.

The attendees were easily able to identify a triple and explain the three parts—

Subject: Identifies an object of interest

Predicate: The relationship between the Subject and Object

Object: Another resource or a literal

Question 2.

The attendees also recognized that a triple with a literal object was still linked data.

During the demo of using the BF Editor to input the serial data a correction was noted. An issuing body in RDA is considered “Other persons, families, and corporate bodies associated with a work” and the guidelines are in RDA 19.3, not in RDA 19.2 as shown in the manual.

After going over the questions, Judy Cannan stressed to the attendees that constant change is in our future.

Submitted by Diane Boehr

Resource 3: Music Score

RIMMF

Melanie Polutta reviewed how to import a bibliographic record from an external OPAC into RIMMF. Any OPAC that permits saving a record in MARC Format can be used. The exported record needs to be in Unicode/UTF-8 form and must have a .mrc extension for RIMMF to recognize.

If you think that an authority record exists in the NAF for the corresponding to the work, download it first into RIMMF. Otherwise, RIMMF will create a work record based on the manifestation record, and the richness contained in the NAR will not be available in the RIMMF record.

Once this step is complete, drag and drop the .mrc record into RIMMF, confirm that you want to import the records, and then review and evaluate the results, moving content among the various entity records as needed. This includes MARC 5XX fields which RIMMF maps to the Manifestation record, even if they more appropriately belong to the Expression.

Note that RIMMF supports multiple iterations of the same data element as needed. Thus, separate Statements of Responsibility go in different lines, and so do extent statements representing a score and a set of parts.

BIBFRAME

Paul Frank reviewed the same resource described using BIBFRAME. This is a complex example, but it reflects the kinds of resources catalogers actually encounter.

Because the BIBFRAME pilot was not a production environment, it was possible to experiment to see what did and did not work in the BIBFRAME 1.0 vocabulary, especially in relation to describing music resources. One major challenge is that BIBFRAME collapses the RDA/FRBR entities of work and expression into a single BIBFRAME work. So, when describing this arrangement, BIBFRAME treats it as two works (the original, and the arrangement). Another problem is that RDA/FRAD currently puts medium of performance at the work, not the expression level. This makes it difficult to properly identify the medium of performance for various arrangements – a critical piece of information for users. This mapping is likely to change with the FRBR Library Reference Model, which proposes putting this element at the expression level in all cases. For the purposes of the pilot, medium of performance was available at the RDA expression/BIBFRAME work level. Paul created an input form to support entering the complex information that can currently be accommodated in the MARC 382 field, and this will create very specific linked data triples.

Note that the BIBFRAME editor will provide drop down lists from various controlled vocabularies, when available. These can be further edited, if the prompted vocabulary doesn’t match the preferred term.

In the visualizations, some triples have the predicate “ and some have “ – in the latter case, these terms were not in BIBFRAME Vocabulary 1.0, and were identified during the pilot as areas where the BIBFRAME vocabulary could be enhanced.

Questions:

  • Why does the manual recommend that RIMMF create an Expression record for the arrangement of Stravinsky’s Rite of spring instead of having the cataloger bring in the existing NAR Stravinsky, Igor, 1882-1971.Vesnasvi︠a︡shchennai︠a︡; arranged?
    A discussion ensued about the benefits and drawbacks of having an undifferentiated record representing multiple expressions.
  • Benefits: much quicker for the cataloger; brings all (arranged) expressions together without having to create a lengthy authorized access point. Note that this is the current PCC policy.
  • Drawbacks: cannot specify date of expression, instrumentation, arranger, and/or other elements unique to a particular expression.
  • Why did the manual suggest removing the San Francisco Symphony Orchestra as a sponsoring body?
    The actual sponsor is the San Francisco Symphony’s Keeping Score educational outreach program, and that’s not currently established in the NAF. Omitting this relationship is cataloger’s judgment.
  • In a linked data environment, errors (assertions in linked data triples) never really go away. This includes inferred dates for works and expressions that could be “corrected” later. What is the impact on our data (and our users) if incorrect statements persist in some way?
  • Is the level of granularity in RDA really needed? How important is it to retain this depth in a linked data environment?
  • How would you map the RDA/FRBR Group 1 entities represented in the RIMMF output for this resource to the corresponding BIBFRAME “entities”? The discrepancy in mapping causes problems, but we have to be able to make this work. This is the one area where we will have the most difficulty, and we need to explore this further.
  • Did the approach taken in entering medium of performance data in BIBFRAME vs. RIMMF cause any problems in terms of linked data? RIMMF is rigid, since it is designed as a visualization of RDA, for the purposes of training and testing. MARC obviously offers greater flexibility, and any MARC successor also needs to allow for bringing information into our “records” that go beyond just RDA elements.

Submitted by Kathy Glennan

Resource 4: MP3 Download

Melanie Polutta showed the cloning technique in RIMFF. She noted the possible complexity of expression relationships.

Paul Frank showed the same resource in BIBFRAME.

Question: how important is a structured form of dates? Discussion included transcribed vs. recorded, various date formats and calendars.

  • Melanie: display is a different issue from how the data for the date are recorded.
  • Another person pointed out that you may not always have the option to record machine actionable data. You need to have options.

Question: Should we be limited to one ontology?

  • Consensus was no.
  • This means giving up control.
  • Is this good for the user? Principle of user convenience.
  • What is PCC's role in this?
  • "Ontology" doesn't necessarily mean "useful". The Library community will need to decide which ontologies are useful for our users.
  • There are a lot of vocabularies out there that are not really "ontologies." PCC needs to narrow its own pool of trusted ontologies.
  • The needs of specialized users need to be accommodated.

Question: What are some reasons for libraries to transition to new technologies?

  • The precision of data is better.
  • Tech services will be able to communicate better with IT and data consumers
  • Replication of data will be easier, avoiding duplicative work.

Question: Is everyone's library discussing linked data?

  • Not all
  • Linked data is being talked about in other contexts than cataloging.

Submitted by Robert Maxwell