Thank you, Library of Congress and Zepheira, for the hard work, and for
publishing a first draft model of a possible post-MARC bibliographic
data transport framework.
I understand that Bibframe's "Work" is consisting of one or more
abstract "core entities" that are created during the process of library
cataloging, by strictly following cataloging rules.
Other entities, which are attached to the cataloged entity, but may be
outside the scope of library cataloging rules, are declared as
The current situation is, MARC records contain a mixture about both
Works and Annotations.
So, from my understanding, the tasks the Bibframe model suggests could
be described roughly by
- identify and extract all the library catalog "core entities" in each
- drop/rephrase odd structures like URLs of online resources in
catalogs, as they are entered as text strings into catalogs, not as
actionable links to resources
- identify other entities in MARC records, or even entities not yet
attached to MARC records, as well as links and relationships to other
entities with external responsibilities
- classify and deduplicate the collected entities, if they represent
resources in the library domain, or if they represent non-library
resources (like access to remote resources)
- assign proper URIs to each entity, or, if not feasible, assign "rich
URL" descriptions to access resources properly
- link the "other entities" to the "library catalog core entities" by
the help of Bibframe
- package and encode the result - a linked entity set - by the help of
future serializing/encoding rules of Bibframe
Is "Bibframe Work" a good choice? I would prefer something like library
domain "core entity", because the term "Work" conflicts with given terms
specific in library catalog rule terminology. I understand Bibframe
"Works" and FRBR/RDA "Works" are not the same.
And is "Bibframe Annotation" a good choice? I would prefer something
like "attachment" or "attached entities". Nevertheless, it would be nice
to find what exactly a "Bibframe work" and a "Bibframe annotation"
constitutes. Maybe "Bibframe work" is always part of the domain of a
library, where "Bibframe annotations" are always part of an open world,
without an exact definition?
I believe there should be no specific preconditions how to attach
arbitrary online resources to a given Bibframe'd library catalog entity.
Creating an URI for identification of the resource, like it is defined
by Linked Data/RDF, and naming the relationship should be sufficient.
I would love to see more references to Linked Data/RDF in Bibframe.
Bibframe should contribute to the problem of solving the library item
"best access" challenge by the public. That is, will Bibframe be able to
support the public access to a given work in a library by representing
information about locations or about the licensing of remote-accessible
Here in Germany, the library union catalogs have the prominent task of
indicating a list of holdings for a given work, enabling services like
My hope is, the "Holding annotations" in Bibframe could serve well at a
global level for library holding exchange, by compiling incrementally
all the holding information of the libraries using Bibframe, and even
more, by embedding information about the remote access to library or
non-library items of a work beside the traditional library item holdings
data of ISO 20775.
For "Holding annotations", I use for my own work a Linked Data/RDF
structure I call "Library Items and Access". It joins the information
about physical and electronic access to library items by using service
descriptors. Each service a library can offer is defined by the
identifier of the library (ISIL), the responsible organization for the
specific item offering (ISIL or non-ISIL), which represent agents like
libraries, publishers, or consortia, and a service type (loan, direct
delivery, inter-library loan, copy, digitization, reference only usage,
etc). Each service type is augmented with transport information for
efficient delivery. For electronic delivery, the URL is listed together
with function codes what the purpose of the URL is (to cover automatic
licensed access), and textual descriptions targeted to the public. For
physical transport, call numbers/barcodes and the physical attributes of
the library item are described, for local or postal delivery.
The resulting access information is a deeply nested data structure. It
could be indexed and displayed in an arbitrary online catalog search
system of a library, where the library serves as a single point of
access, also with hints about external library and non-library resources
and how to get it, which is very useful for inter-library loan and
If such access information for a resource-sharing library network could
be encoded in Bibframe, it would be a greatly appreciated. It would be
progress for the benefit of all of us.
Thanks again for your effort, and keep up the good work!