On Mar 11, 2013, at 5:04 PM, Owen Stephens <[log in to unmask]> wrote:
> I don't think there has been any discussion on the list about this yet (sorry if I've missed it)
If you missed it, that makes two of us. But your question relates to an important design decision so thanks for bringing this up. To back up a bit (and ensure shared context) the following is a bit dated, but I think still quite useful:
[[
The BIBFRAME Vocabulary
The BIBFRAME model is defined in RDF.
The RDF vocabulary for the draft BIBFRAME model, to be available shortly, will provide a convenient way of navigating around the RDF model.
The suggested namespace for the BIBFRAME model is http://loc.gov/bibframe/vocab. An RDF/XML serialization of this vocabulary is expected to be available at this location. While the recommendation of a singular namespace is counter to several current Linked Data bibliographic efforts, it is crucial to clarify responsibility and authority behind the schematic framework of BIBFRAME in order to minimize confusion and reduce the complexity of the resulting data formats. It will be the role of the Library’s standards stakeholders to maintain the connections between BIBFRAME model elements and source vocabularies such as Dublin Core, FOAF, SKOS and future, related vocabularies that may be developed to support different aspects of the Library workflow.
Identifier persistence is not a technology issue but rather an organizational, policy, and community one. A persistence policy should be defined stating clearly the persistence and change management mechanisms.
]]
-- http://www.loc.gov/marc/transition/pdf/marcld-report-11-21-2012.pdf
But certainly more background and detailed discussion is warranted. Quickly scanning the growing thread around this topic, it seems that many think so as well ;)
> Much of the practical work in the area of linked data and libraries has made use of common existing vocabularies to publish data. This commonly includes:
>
> DC/DC Terms
> FOAF
> Bibliographic Ontology (aka Bibo)
And thats just a start!
> I think I've seen these in every implementation of linked data representations of library materials. They are often used alongside other existing vocabularies (Bio, frbr, isbd) and often a specifically created vocabulary to deal with elements that fall outside these existing vocabs (e.g. the BNB uses http://www.bl.uk/schemas/bibliographic/blterms#, Libris uses http://libris.kb.se/vocabulary/experimental#)
Our initial linked data analysis identified literally dozens of namespaces used in the Library community reflecting various stages of experimentation. More if you start to broaden analysis to the include museums, archives, galleries, etc.. And even more if you broaden into more of how we might imagine patrons / others adding value to this data. This shouldn't be a surprise to anyone familiar with RDF / Linked Data. There isn't one vocabulary that works (or will work) for everything. ;)
> In light of this it seems notable that so far Bibframe doesn't specify the use of any vocabularies outside bibframe specific elements, and there is no information on how the bibframe elements might map to existing vocabularies.
Good point. Eventually we do intend to work on non-normative transform and mappings from and too other vocabularies, which allows pragmatic reuse. And over time, with usage experience some of those mappings might take on normative force. But certainly this is underspecified and more work in this area is required.
> So I guess my first question is - shouldn't Bibframe reuse existing vocabularies where possible? If there are arguments against this - what are they?
There are many benefits of vocabulary reuse, but as with many things, there are costs as well that need to be carefully considered. As Ross points out, you can't control the fate of vocabularies you don't own. Designing systems that leverage multiple vocabularies managed by various stakeholders is a tricky issue and one that requires careful considerations. There are many reasons why namespaces / vocabulary "drift" over time (404 being a worse case example) and all of these may have an affect on systems. Business acquisitions (anyone recall http://my.netscape.com/rdf/* ?), economic factors, organizational changes, changing social interests, etc. are just a handful of reasons for causing such change, but thinking ahead to infrastructure to support the next 40+ years of Libraries, namespace persistence is a key point to consider when dealing with how best to integrate and invest in vocabulary terms outside of ones community.
We accept a level of risk by basing BIBFRAME on open, royalty free W3 standards because we have confidence in a shared (Web) community's investment in these underpinning vocabularies. Basing BIBFRAME on RDF (and the Web) allows us to more easily weave in additional vocabularies as new requirements, customization and localization demand while still achieving resilience in systems. While BIBFRAME is being built on a framework that supports mix-matching of terms, its taking an appropriately critical view regarding minting new terms vs adopting external ones. Making formal connections between BIBFRAME model terms and other vocabularies terms errs on the side of caution when striking a balance between the stability of BIBFRAME (and an existing community in which it serves) and a web of relationships with other communities. And while I expect this will lead to some duplication of terms, as Karen pointed out, Library terms are generally more precise than other (intentionally broader) community terms. So, while I expect Library’s standards stakeholders to maintain the connections between BIBFRAME model elements and other relevant vocabulary terms, I expect many of these relationships to be "is Refinement Of" ones. ;)
At this point, i'd be remiss in not mentioning the vocabulary is in development, draft, subject to change, etc.
In the meantime, my hope is though this effort we will also collectively advance the discussion of namespace persistence to help ensure weather we map to terms (or mix / match these in various community profiles) the terms in question will continue to exist though out the course of BIBFRAME. There are many examples of stated persistence policies for vocabularies [0][1][2][3]. But I think far more could be done (a simple set of machine readable 'living will' templates might be an example of such a step [4]) to help ensure social and organizational backups.
Again, identifier persistence is not a technology issue but rather an organizational, policy, and community one. A value in BIBFRAME namespace(s) will be proportionate to the community that invests in them as well as the organization commitment to these namespaces and terms.
--
Eric Miller
President, Zepheira "The Art of Data"
http://zepheira.com/ tel:+1.617.395.0229
[0] W3C URI Persistence Policy
- http://www.w3.org/Consortium/Persistence.html
[1] Namespace Policy for the Dublin Core Metadata Initiative (DCMI)
- http://dublincore.org/documents/dcmi-namespace/
[2] DCMI Generic Namespace Policy for RDF Vocabularies
- http://dublincore.org/documents/2011/05/02/dcmi-namespace-generic/
[3] Agreement between DCMI and the FOAF Project
- http://dublincore.org/documents/dcmi-foaf/
[4] I think a simple set of machine readable templates may serve as a very useful means for defining a "living will" for namespaces. Something akin to CC templates that contains...
1) The owner
2) The namespace URI(s)
3) A policy concerning classes of changes that are allowed
4) Designator on institution committed to taking over namespace in the event a current party is (for whatever reason) no longer able to do so.
might be a useful step in formalizing various informal policies that would aid in stakeholder discussions.
|