Hey all,

2 more cents from me. In the Linked Data world, based on URIs and RDF,
two distinct classes of resources are most important:
- information resources (digital documents)
- non-information resources (physical objects and abstract concepts)

Instances of those 2 cannot be conflated. For example, having the same
URI identify a person and the document about that person will lead to
problems sooner rather than later. Fortunately, this problem is easily
solved using proper URI design and giving these instances different
URIs. They can still resolve to the same document, either using
redirects or hash URIs:

I can imagine that in bibliographic data there are further sub-divisions:
- some book as a concept
- specific edition of that book
- specific physical copy of that edition of that book

and so on. Figuring those levels of abstraction might be a modeling
problem, but they do not pose a problem in the conversion to RDF as
long as the URI design is sound.


On Wed, Mar 11, 2015 at 7:06 PM, Ross Singer <[log in to unmask]> wrote:
> I'm actually not attributing anything to dbpedia except its identifiers.
> Dbpedia and its data model are a mess and I wouldn't suggest using its
> assertions except in select conditions.
> Its identifiers (and what we're really saying here is wikipedia's
> identifiers) *are* very useful, and can be useful while ignoring the rest of
> the content and data therein.
> You're absolutely right about relying on dbpedia or many of the things it
> links to, currently. Much of it is hobbyist, academic exercises, or
> orphaned. It's not necessarily unexpected for a data model that requires
> lots of data to show its value to have a lot of unofficial implementations
> of data sets to help prime the pump.
> I think the question becomes less "what will we link to if we move to RDF"
> and more "does the linked data pattern hold value and would RDF be a
> reasonable avenue to help us migrate to it"?
> MARC wasn't created for the things it could immediately do for libraries, it
> was for the possibilities it could bring: namely automating the printing of
> catalog cards, since the current methods couldn't keep up with proliferation
> of resources and machines and networks (albeit, generally physical networks,
> in this case) could help us be more efficient. It certainly changed the
> duties of the cataloger, but didn't change their importance.
> We're at the exactly the same crossroads now.
> -Ross.
> On Wednesday, March 11, 2015, Charles Pennell <[log in to unmask]> wrote:
>> I suspect that you are attributing more power to DBPedia than it
>> possesses, having quite a bit of experience working with it in conjunction
>> with NCSU's nascent Organization Name linked data project
>> ( I will grant you that one can find
>> significant information about the descriptive attributes of any given
>> resource, in some cases even including about a particular edition of a work,
>> if it is well enough known to have been written about and linked to.  And
>> isn't behind a pay-for-view firewall that you haven't paid for. Or
>> out-of-date.  Over time, there will obviously be even more descriptive data
>> available. But there is not so much describing the actual contents of the
>> work and even less by unbiased agents, or at least those who don't stand to
>> profit by directing people to a particular resource. Trust in our
>> objectivity was one of the principle advantages of library metadata,
>> insufficient though the metadata might seem in hindsight. One need only to
>> listen to politicians or to watch Fox News to see where naivete on our data
>> sources gets us.
>> Just to note: None of this is an argument against linked data. Linked data
>> has already shown its ability to expose us to a lot more aspects of any
>> given topic or entity than we've previously had exposure to.  But just like
>> our historic metadata, it has its own shortcomings.  And it also has the
>> potential to lead us down a lot more rabbit holes occupied by spurious and
>> perhaps even false information.
>>   Charley
>> On Tue, Mar 10, 2015 at 7:03 PM, Ross Singer <[log in to unmask]>
>> wrote:
>>> I'm not sure what you're asking here.  You mentioned that we can point an
>>> author or publisher to VIAF or DBPedia, but why didn't you mention that we
>>> could link the work?  Or the edition?
>>> And then if that link to the work is linked to dbpedia, and dbpedia links
>>> to another dataset that links the works that were inspired or adapted from
>>> the original work, or lists the places where the action takes place in the
>>> original work, or citations to/from this work, or any countless numbers of
>>> possibly useful or useless or somewhere in between links between data.
>>> You're right that we likely won't get access to the fulltext although I'm
>>> not convinced that would be terribly useful, anyway.  But we are quite
>>> likely to be able to have access and contribute to a lot more metadata to
>>> find the relationships *between* resources, which, in turn, opens up
>>> discovery.
>>> -Ross.
>>> On Tue, Mar 10, 2015 at 5:44 PM Charles Pennell <[log in to unmask]>
>>> wrote:
>>>> If the base descriptive data lacks sufficient elements to link out to
>>>> other data sources, how does RDF enhance it?  Sure, you might have an author
>>>> or publisher to link out to VIAF or DBPedia, or a single subject heading to
>>>> link to, but how is that going to provide you with more
>>>> information on the contents of the actual resource being described?  Short
>>>> of linking to full-text, which is not going to happen for anything published
>>>> after 1928 (except for purchased content, in the case of more recent
>>>> materials), what will linked data provide us with to work with our existing
>>>> print resources?  I'm not arguing that MARC is inherently superior or
>>>> inferior to Bibframe, only that in either case we are subject to the same
>>>> limitations in accessing the contents of our resources.
>>>>    Charley
>>>> On Tue, Mar 10, 2015 at 11:34 AM, Ross Singer <[log in to unmask]>
>>>> wrote:
>>>>> It doesn't have to be there if it links to other things.
>>>>> The problem is with MARC is that that record (and each copy of it in
>>>>> each place) would need to be updated with every enhancement.  How would the
>>>>> records incrementally improve?
>>>>> With BIBFRAME and RDF, the data *doesn't* have to be there.  That's the
>>>>> whole point of it.  But by identifying it and the resources in it, we can
>>>>> make inferences from other data.  And since this is the design from the
>>>>> start, it's not shoehorning onto a data format that isn't particularly well
>>>>> suited for it.
>>>>> -Ross.
>>>>> On Tue, Mar 10, 2015 at 2:55 PM Charles Pennell <[log in to unmask]>
>>>>> wrote:
>>>>>> Of course you do realize that if the data isn't there in our MARC
>>>>>> records, it surely isn't going to be in our Bibframe/rdf records either.
>>>>>> Providing buckets for data isn't the same as filling them, and the money
>>>>>> will never be there for retroactively bringing historic records up to
>>>>>> standards.  Really, it is no longer there for bringing our current records
>>>>>> up to standard either, as can be witnessed in the proliferation of
>>>>>> vendor-supplied brief and substandard records we are all ingesting to manage
>>>>>> our e-resources.
>>>>>>   Charley
>>>>>> On Tue, Mar 10, 2015 at 10:41 AM, Ross Singer <[log in to unmask]>
>>>>>> wrote:
>>>>>>> Robin,
>>>>>>> I am happy to be corrected, but I would need to see some examples, in
>>>>>>> MARC, that satisfy the scenarios I laid forth.
>>>>>>> I deal with lots and lots of MARC data and none of it can answer
>>>>>>> those kinds of questions, and, worse, a lot of it isn't even possible to
>>>>>>> link to other data where I could answer those questions.
>>>>>>> Remember, when we talk about MARC data, we can't just talk about
>>>>>>> what's *possible* in MARC21, but what actually is in the records we have:
>>>>>>> since each record is a discrete standalone document, if the data isn't
>>>>>>> there, it's nearly impossible to improve upon that.
>>>>>>> Or, basically, what Karen just said.
>>>>>>> -Ross.
>>>>>>> On Tue, Mar 10, 2015 at 2:04 PM Wendler, Robin King
>>>>>>> <[log in to unmask]> wrote:
>>>>>>>> My reaction was the same as Cecilia’s.  MARC can carry that
>>>>>>>> information. There has been a vicious circle, though, because without
>>>>>>>> systems building functionality to use the metadata, libraries have been
>>>>>>>> unlikely to invest in creating it. It’s been increasingly difficult to
>>>>>>>> justify this level of detail in cataloging to administrators. We can lay
>>>>>>>> plenty of sins at MARC’s door, but not these particular ones.
>>>>>>>> Robin
>>>>>>>> Robin Wendler
>>>>>>>> Library Technology Services
>>>>>>>> Harvard University
>>>>>>>> 90 Mt. Auburn St.
>>>>>>>> Cambridge, MA 02138
>>>>>>>> 617-495-3724
>>>>>>>> [log in to unmask]
>>>>>>>> From: Bibliographic Framework Transition Initiative Forum
>>>>>>>> [mailto:[log in to unmask]] On Behalf Of Cecilia M. Preston
>>>>>>>> Sent: Tuesday, March 10, 2015 9:38 AM
>>>>>>>> To: [log in to unmask]
>>>>>>>> Subject: Re: [BIBFRAME] Linked data
>>>>>>>> Ross,
>>>>>>>> My first reaction to many of these questions about what MARC has not
>>>>>>>> or could not do, has been talk to your ILS vendor.  Much of the data
>>>>>>>> embedded in a MARC record was never indexed by the ILS folks because ‘no one
>>>>>>>> was interested in it’ or what was generally referred to in the ancient days
>>>>>>>> of Z39.50 ATS (Author, Title, Subject) was all the patron/user/client is
>>>>>>>> interested in.  When I asked the vendor for my local system at ALA why I
>>>>>>>> could not get access to some information I knew was in a MARC record the
>>>>>>>> answer was simply ‘we can index that if they want to $$$ for it’  Not
>>>>>>>> something I think many institutions have the funding to do for me.
>>>>>>>> -Cecilia
>>>>>>>> On Mar 5, 2015, at 6:25 PM, Ross Singer <[log in to unmask]>
>>>>>>>> wrote:
>>>>>>>> Counterpoint: if libraries can do "anything they want" with their
>>>>>>>> data and have had 40+ years to do so, why haven't they done anything new or
>>>>>>>> interesting with it for the past 20?
>>>>>>>> How, with my MARC records alone, do I let people know that they
>>>>>>>> might be interested in "Clueless" if they're looking at "Sense and
>>>>>>>> Sensibility"? How do I find every Raymond Carver short story in the
>>>>>>>> collection? The albums that Levon Helm contributed to? How can I find every
>>>>>>>> introduction by Carl Sagan?  What do we have that cites them?
>>>>>>>> How, with my MARC records alone, can I definitively limit only to
>>>>>>>> ebooks? What has been published in the West Midlands?
>>>>>>>> You *could* make a 3-D day-glo print of a MARC record, I suppose -
>>>>>>>> but that seems like exactly the sort of tone deaf navel gazing that has
>>>>>>>> rendered our systems and interfaces more and more irrelevant to our users.
>>>>>>>> -Ross.
>>>>>>>> On Thursday, March 5, 2015, J. McRee Elrod <[log in to unmask]> wrote:
>>>>>>>> Forwarded by permission of James Weinheimer:
>>>>>>>>   There are some points to keep in mind when considering linked
>>>>>>>>   data/semantic web. The new formats (, Bibframe) are
>>>>>>>> *not*
>>>>>>>>   there for libraries to be able to do new and wonderful things with
>>>>>>>> their
>>>>>>>>   own data. Why? Because libraries already understand and control
>>>>>>>> all of
>>>>>>>>   that data. Right now, so long as we have XML formats (and we have
>>>>>>>> that
>>>>>>>>   now with MARCXML) we can do *anything* we want with the data.
>>>>>>>> MARCXML is
>>>>>>>>   not perfect, but it is still XML and that means: librarians can
>>>>>>>> search
>>>>>>>>   that data however we want, manipulate it however we want,
>>>>>>>> transform it
>>>>>>>>   however we want, sort it however we want and display it however we
>>>>>>>> want.
>>>>>>>>   If we want to search by the fiction code in the fixed fields and
>>>>>>>> sort by
>>>>>>>>   number of pages or by 100/700$q we can. We can print out reams of
>>>>>>>> entire
>>>>>>>>   records, or any bits and pieces of them we could want, collate
>>>>>>>> them in
>>>>>>>>   any number of ways (or not), and print them out on 3D printers in
>>>>>>>>   day-glow colors, display them with laser beams on the moon or work
>>>>>>>> with
>>>>>>>>   them in the virtual reality "wearable technology". We can do all
>>>>>>>> of that
>>>>>>>>   and more *right now* if we wanted. We've been able to do it for a
>>>>>>>> long
>>>>>>>>   time. We don't need or Bibframe to enhance our own
>>>>>>>>   capabilities because we can do anything with our own data now.
>>>>>>>>   So, who is and Bibframe for? Non-librarians, i.e. for
>>>>>>>> people
>>>>>>>>   who neither understand nor control our data. Libraries will allow
>>>>>>>> others
>>>>>>>>   to work with our data in ways that they can understand a bit more
>>>>>>>> than
>>>>>>>>   MARC. Non-librarians cannot be expected to understand 240$k or
>>>>>>>> 700$q,
>>>>>>>>   but with or Bibframe, it is supposed to be easier for
>>>>>>>>   them--although it still won't be easy. Nevertheless, they will be
>>>>>>>> able
>>>>>>>>   to take our data and do with it as they will as they cannot do now
>>>>>>>> with
>>>>>>>>   our MARC/ISO2709 records.
>>>>>>>>   With Bibframe and people will be able to merge it with
>>>>>>>> other
>>>>>>>>   parts of the linked data universe (oops! Not Freebase or dbpedia.
>>>>>>>>   They'll have to go to Wikidata! Wonder how long that will last!)
>>>>>>>> or with
>>>>>>>>   all kinds of web APIs (see
>>>>>>>> that
>>>>>>>>   can create mashups. (I still think this video gives the best
>>>>>>>> description
>>>>>>>>   of a mashup: What is a mashup? - ZDNet. Here too is a
>>>>>>>>   list of some of the web apis
>>>>>>>> Web programmers can
>>>>>>>> then
>>>>>>>>   put these things together to create something absolutely new, e.g.
>>>>>>>> bring
>>>>>>>>   together library data with ebay so that people can see if
>>>>>>>> something on
>>>>>>>>   ebay is available in the library or vice versa. But remember that
>>>>>>>> those
>>>>>>>>   web programmers will also be able to manipulate our data as much
>>>>>>>> as we
>>>>>>>>   can, so the final product they create may look and work completely
>>>>>>>>   differently than we would imagine, or that we would like. As a
>>>>>>>> result,
>>>>>>>>   libraries and catalogers will lose the control of their data that
>>>>>>>> they
>>>>>>>>   have always enjoyed. For better or worse, that is a necessary
>>>>>>>>   consequence of sharing your data.
>>>>>>>>   Then comes what are--I think--the two major questions of linked
>>>>>>>> data for
>>>>>>>>   libraries. First is: OK. We add the links, but what do we link
>>>>>>>> *to*?
>>>>>>>>   Will linking into appeal to the public? I personally
>>>>>>>> don't
>>>>>>>>   think so since there is so little there, other than the
>>>>>>>> traditional
>>>>>>>>   syndetic structures found in our traditional catalogs (i.e. the
>>>>>>>> UF, BT,
>>>>>>>>   NT, RT for subjects, the earlier/later names of corporate bodies
>>>>>>>> and
>>>>>>>>   series, the other names of people). This is not what people think
>>>>>>>> of
>>>>>>>>   when they think of the advantages of linked data. While those
>>>>>>>> things may
>>>>>>>>   be nice for us, I don't know if that will be so appealing to the
>>>>>>>> public.
>>>>>>>>   If it is to become appealing to the public, somebody somewhere
>>>>>>>> will have
>>>>>>>>   to do a lot of work to make them appealing.
>>>>>>>>   Concerning VIAF, it's nice to know the authorized forms in Hebrew,
>>>>>>>>   French, Italian, and so on, but again, is that so appealing to the
>>>>>>>>   *public*? It may be, but that remains to be proven.
>>>>>>>>   Second, there is no guarantee at all that anyone will actually do
>>>>>>>>   anything with our data. While I certainly hope so, there are no
>>>>>>>>   guarantees that anybody will do anything with our data. It could
>>>>>>>> just
>>>>>>>>   sit and go unused.
>>>>>>>>   It's interesting to note that the LC book
>>>>>>>>   catalog in this format has been in the Internet Archive for awhile
>>>>>>>> now
>>>>>>>>   ( but I
>>>>>>>> haven't
>>>>>>>>   heard that any developers have used it.
>>>>>>>>   I want again to emphasize that libraries should go into linked
>>>>>>>> data, but
>>>>>>>>   when we do so, there will probably be more question marks than
>>>>>>>>   exclamation points. Just as when a couple is expecting a baby and
>>>>>>>> they
>>>>>>>>   experience pregnancy: at least when I experienced it, I imagined
>>>>>>>> that
>>>>>>>>   the birth of my son would be an end of the pregnancy. But
>>>>>>>> suddenly, I
>>>>>>>>   had a crying baby on my hands! Linked data will be similar: it
>>>>>>>> will be a
>>>>>>>>   beginning and not an end.
>>>>>>>>   James Weinheimer [log in to unmask] First Thus
>>>>>>>> First Thus Facebook Page
>>>>>>>> Cooperative Cataloging Rules
>>>>>>>>   opencatalogingrules Cataloging Matters
>>>>>>>>   Podcasts
>>>>>>>> [delay
>>>>>>>>   +30 days]
>>>>>>>>   --
>>>>>>>> --
>>>>>>>> James Weinheimer [log in to unmask] First Thus
>>>>>>>> First Thus Facebook Page
>>>>>>>> Cooperative Cataloging Rules
>>>>>>>> opencatalogingrules Cataloging Matters
>>>>>>>> Podcasts
>>>>>>>> [delay
>>>>>>>> +30 days]
>>>>>> --
>>>>>> Charley Pennell
>>>>>> Principal Cataloger
>>>>>> NCSU Libraries
>>>>>> North Carolina State University
>>>> --
>>>> Charley Pennell
>>>> Principal Cataloger
>>>> NCSU Libraries
>>>> North Carolina State University
>> --
>> Charley Pennell
>> Principal Cataloger
>> NCSU Libraries
>> North Carolina State University