The fault is not in our cataloging code but in ourselves. RDA treats frequency as a distinct element (2.14), and requires a note only when an appropriate term is unavailable. The terms at RDA 18.104.22.168 correspond to the codes available in the MARC bibliographic format (008/18), and even more elaborate coded frequency data can be stored in the MARC holdings format. Rather, the proliferation of notes is often a result of implementation decisions, and these are often informed by the perceived capabilities of existing machine systems. For example, there is no reason that codes could not have been developed for the plethora of relationship designators in RDA appendices I-L, but instead we supply the terms verbatim (and in English) from RDA. Likewise with the carrier, media, and content types; for the first two, coded equivalents were already in existence for the most part, so the text recorded (in English) in fields 337 and 338 is often redundant with coded values already present in 007/00-01. I, for one, would be happy to see notes yield to coded data whenever possible, if only to save keystrokes.
Associate Director, Library Assessment and Technical Services
National University Library
9393 Lightwave Avenue
San Diego, California 92123-1447
+1 858 541 7920 (voice)
+1 858 541 7997 (fax)
From: Bibliographic Framework Transition Initiative Forum [mailto:[log in to unmask]] On Behalf Of Karen Coyle
Sent: Monday, September 15, 2014 11:33 PM
To: [log in to unmask]
Subject: Re: [BIBFRAME] Literal Properties vs Notes
Frequency is a good example of the mess that we have today in AACR-type cataloging. There is a way to code frequency as a calculable value in the MARC holdings record. That should suffice and there should be no need for a note field if the value is provided. But notes are "notes" because they are uncontrolled strings created by the cataloger. Where other data in the record are either transcribed strings or controlled headings, notes are neither. But the notes ARE required by the cataloging rules.
This is evidence, to me, of the gap between the cataloging rules and the actual practice of creating machine-readable catalog records. AACR does not recognize that coded data (e.g. MARC fixed fields) exists.
Many notes repeat information that could be encoded elsewhere, but because the note is what is required by the cataloging rules and also displays in the catalog, the tendency has been to provide the note but often not to provide the actionable data element. Obviously, it would be a mistake to carry forward this practice, and instead the actionable data element must be the primary source of data, from which user-friendly notes can be derived if needed. That "if needed" part is also something we should think about, because in fact in many system displays notes are not included, so catalog users rarely see them.
Because notes have been favored over actionable data, there is a whole host of information that is 1) not usable for any automated functions and 2) rarely seen by users. Surely this is a waste of cataloger time, and a disservice to our users.
Quoting Tim Thompson <[log in to unmask]>:
> Having a bf:Note class makes sense to me. The current approach seems
> exhaustive enough to be cumbersome, but probably not exhaustive enough
> to capture the full range of possibilities in the source data. Not all
> notes come from 5XX fields. Here is a sample marc2bibframe conversion
> of a record for a serial:
> MARC: http://bibframe.org/resources/Jqc1410365115/marcxml.xml
> BF: http://bibframe.org/resources/Jqc1410365115/bibframe.rdf
> Here, bf:frequencyNote maps to the 310 field (Current Publication
> Frequency). Unfortunately, it also maps to the 321 field (Former
> Publication Frequency). This would seem to be a not insignificant loss
> of information. 5XX fields that are distinct in MARC are mapped to
> generic bf:note properties (515, 588). bf:frequency doesn't appear,
> but maybe it was meant to correspond to the 008 fix field for
> continuing resources, which also has a value for frequency (position
> 18). The need for two distinct properties remains unclear.
> In short, a bf:Note class with bf:noteType values might provide
> greater flexibility and preserve more of the original semantics.
> Tim A. Thompson
> Metadata Librarian (Spanish/Portuguese Specialty) Princeton University
> On Tue, Sep 9, 2014 at 5:38 PM, Robert Sanderson <[log in to unmask]>
>> Y'all ready for this? ;) 
>> When is a literal property a 'somethingNote' and when is it just a
>> I assume (lacking previously mentioned MARC to BibFrame mapping
>> document) that all of the Notes come from 5XX fields, which seems
>> like something that could easily be rationalized along with some of
>> the other properties, again assuming they're not 5XX and hence didn't get the Note moniker.
>> For example, these two look ... well ... identical:
>> frequency: Intervals at which the issues or parts of a serial or the
>> updates to an integrating resource are issued.
>> frequencyNote: Current or former publication frequency of a resource.
>> Current notes are:
>> note (!)
>> And the following seem like they're intended to be "notes" in the
>> more generic sense of added description by a cataloguer or other:
>> providerStatement -- or are "Statements" transcriptions from the
>> contentAccessibility (though c.f. schema.org/accessibilityFeature)
>> Given the discussion regarding assigners of URIs being important, it
>> seems that creators of notes would be also important? And thus Notes
>> could be their own class, bf:Note, with properties including value,
>> assigner, type and so forth.
>>  https://www.youtube.com/watch?v=avcS0aYJ2a8 Warning: seizure
>> inducing flashing, terrible animation, poppy 90s music, ...
>> Rob Sanderson
>> Technology Collaboration Facilitator
>> Digital Library Systems and Services
>> Stanford, CA 94305