> If graph theory really was the mathematical idea that animated RDF, the ability to define resource description subgraphs should(!) have been implemented in RDF and propagated to RDA and to BIBFRAME.
I think that ability has begun to appear. The notion of named graphs (now packed into the notion of "RDF dataset") can be used to support "extensively" defined subgraphs, and various query languages can be used to support "intensively" defined subgraphs. There are other kinds of subgraphs defined by some recursive algorithm (e.g. RDF Molecules, Concise Bounded Descriptions, etc.) that may help support functionality like provenance or authorization. There's surely a long way to go (especially when we want to talk about Linked Data), and I would never claim that graph theory is the soul of RDF, but then, the fact that tree automata were not the motivation for XML didn't make them turn out to be any less useful.
Personally, I'm concerned that the notion of subgraph be available to the library community because of something given up with the loss of the bibliographic record as a universal container: a natural unit of work. Operating in a sea of atomic assertions, some way to package up work to be done or shared seems very important to me, and a flexible and powerful ability to define subgraphs seems to me to be at the heart of that facility.
---
A. Soroka
The University of Virginia Library
On Jun 27, 2014, at 11:02 AM, "Murray, Ronald" <[log in to unmask]> wrote:
> Subgraphs are to graphs as subsets are to sets.
>
> If RDF and RDA mavens really intended to treat resource descriptions as
> graphs, they would have known that the ability to define subgraphs is a
> key element of graph theory. (In the first textbook on the topic, Dčnes
> König did so nearly immediately after defining the mathematical object
> known as a graph.)
>
> If graph theory really was the mathematical idea that animated RDF, the
> ability to define resource description subgraphs should(!) have been
> implemented in RDF and propagated to RDA and to BIBFRAME.
>
> Specialized profiles of the type mentioned below can then be understood as
> subgraphs of more comprehensive (though not necessarily ever directly
> encountered) resource description graphs.
>
> The need for resource description subgraphs might be understood as a
> function of the "resolution" to which a resource is "viewed" by a
> describing community. Small wonder that our community has such a need -
> look at our previous implementations. No surprise that many other,
> technically influential, describing communities do not.
>
> Ron Murray
>
>
>
> On 6/26/14 5:24 PM, "J. McRee Elrod" <[log in to unmask]> wrote:
>
>> Karen said:
>>
>>
>>> This is an interesting vision from the cataloger perspective. If I read
>>> you correctly, in a distributed network, for the same item there could
>>> be "records" (profiles or graphs of data) that are suitable for public
>>> libraries, ones that are suitable to specialized libraries, others
>>> suitable to national libraries, or archives, etc.
>>
>> What would a vendor do who has clients in all of the categories above?
>> What does a vendor do who can afford to have only one version of data
>> per manifestation in its files for all clients?
>>
>> Our bottom line depends on the reuse of records.
>>
>> We already have this problem with phrases in different languages (if
>> one does not use the ISBD abbreviations as codes). This would be even
>> more difficult. Perhaps I will just have to shut up shop? There
>> seems to be no place for such services as SLC in this brave new
>> bibliographic world.
>>
>> It has been said that only the needs of larger libraries with inhouse
>> IT ability are being served. That may be correct. I don't see how a
>> small library with no professional staff would be able to function in
>> such an environment.
>>
>>
>> __ __ J. McRee (Mac) Elrod ([log in to unmask])
>> {__ | / Special Libraries Cataloguing HTTP://www.slc.bc.ca/
>> ___} |__ \__________________________________________________________
|
