Re: Disjointedness of FRBR classes

Quoting Tom Baker <tbaker@tbaker.de>:


>>                                          The RDA folks seem
>> surprised when some of us demonstrate to them various ways that
>> they've painted themselves into a corner.
>
> Officially?  Was there or will there be a public comment period for the RDF
> expression?  That could perhaps help surface these issues.

No, there hasn't been a public comment period or any kind of external  
review. All of the info is openly available online, but it's pretty  
hard to dig through hundreds of properties to try to understand the  
decisions that were made when creating the vocabulary. For RDA we did  
publish an article [1], but it was more of a 'this is what we did'  
rather than a call for comments. Even if it had been, I think getting  
feedback on anything this technical is going to take some concentrated  
effort, something beyond "take a look and tell us what you think." The  
PREMIS [2] folks recently published their vocabulary in RDF and I  
tried to look at it but it's overwhelming. If nothing else, developers  
could help if they made a list of things they struggled with as a way  
to start a discussion. But what we really need is some best practices  
that we can rally around and compare these implementations to.

>
>>                                           This tells me that they
>> simply do not understand the implication of formal models, and were
>> assuming that you could "fudge" if you needed to. At the same time,
>> they argue for precision and rigidity.
>
> What is the actual rationale, or requirement, for precision?  If the  
> answer has
> to do with "quality control for cataloging", then perhaps there are  
> other ways
> to achieve this.

As I understand it, the rationale is based on the wide sharing of data  
-- and not just that there is some sharing of data elements but  
libraries share entire records as a way to avoid duplication of  
effort. They also spend a lot of time merging records creating for the  
same things but in slightly different contexts. However, there has  
been no real study of how much 'sameness' is necessary for these  
functions, so it may be the case that 'same' does not mean every data  
element needs to be the same. That said, library cataloging has been  
designed to achieve 'same' across institutions, and there is criticism  
of RDA for allowing too much latitude for decision-making at the local  
level.

Now that there is talk of cloud computing for libraries there may be  
fewer local record modifications being made, and 'same' will mean  
accepting what is in the cloud. That does not make catalogers happy  
because they have a strong sense of what is right and what is wrong,  
and many records created outside of their particular institution turn  
up in the 'wrong' category.


>
> I wonder if there is a middle position -- e.g., properties with domains and
> ranges that are not disjoint with each other.  Have these questions been
> discussed, and on a citable mailing list?

Not that I know of. In fact, except for DC-RDA I don't know of lists  
that have discussions at that level of technical detail. As Diane H  
will attest (this was the topic of her talk at Code4Lib earlier this  
year), the library techies hang out in their lists, and the catalogers  
hang out in a separate set of lists, and it's a real struggle to find  
any twain where they can meet. The differences in their viewpoints and  
knowledge is notable.


>
> Are there any definitions _not_ open to interpretation?  Semantic Web
> colleagues I know who know something about FRBR tend to associate it with
> lively discussions about interpretation.

There've been lively discussions but in the end the view of the FRBR  
review committee is the official one. There are some dissenting  
sub-communities, notably music and serials catalogers. The European  
RDA Users' Group recently announced that they would like to see some  
changes in how RDA interprets FRBR, but I was told off-list that JSC  
considers their interpretation of FRBR to be "wrong."


>
> Perhaps the RDF expressions could be put up for review, and the Semantic Web
> community could be invited to have a look at them from a formal  
> point of view.
> I'd expect one would get ALOT of comments on the disjointness of  
> those classes.

Maybe this is a first or second activity of the LLD community group,  
if it gets started. I would guess that there are folks working with  
bibliographic data who would be disappointed if they couldn't link  
relatively easily to library resources, so there should be a group of  
non-library folks with an interest in making sure this can be done.

>
> Does anyone know if the RDA properties have already started to be  
> used in data?
> Does the status "published" mean they are intended already to be used in
> production data?  It sounds like there are still quite a few open  
> issues to be
> considered.

The RDA properties do not yet have status 'published.' There are a few  
value lists that have been published, but the majority is still  
'proposed.' I've used RDA properties in the Open Library RDF. I have  
seen them used in library RDF in the LLD data sets. I believe that the  
XC project uses some RDA properties.[3] No one that I know is using  
RDA heavily - most are selecting only a few properties that aren't  
available elsewhere. And I would bet that many of them are not  
compliant with RDA's FRBR domains and ranges, although the registered  
RDA properties [4] are not as strictly defined (yet) as FRBRer.[5]

kc

[1] http://dlib.org/dlib/january10/hillmann/01hillmann.html
[2] http://annotation.lanl.gov/premis/
[3] http://extensiblecatalog.org/
[4] http://metadataregistry.org/rdabrowse.htm
[5] http://metadataregistry.org/schema/show/id/5.html

>
> Tom
>
> --
> Tom Baker <tom@tombaker.org>
>



-- 
Karen Coyle
kcoyle@kcoyle.net http://kcoyle.net
ph: 1-510-540-7596
m: 1-510-435-8234
skype: kcoylenet

Received on Monday, 24 October 2011 02:25:58 UTC