Re: Redefine and reuse?

On Tue, Jul 23, 2013 at 02:18:02PM +0200, Dan Brickley wrote:
> On 23 July 2013 00:11, Karen Coyle <kcoyle@kcoyle.net> wrote:
> > Hello. The schemaBibEx group is working to fill in some missing properties
> > needed for bibliographic data, as well as some that are needed for
> > library-specific data. Libraries have long existed in a data silo that has
> > kept us apart from other information communities, so we are quite sensitive
> > about the importance of reusing elements that have a broader context.
> >
> > To this end, we have been looking at properties like "Offer" and "sku" to
> > handle what libraries have to offer: lending of materials, online access,
> > placing of requests (holds). What we are finding, though, is that in many
> > cases the definitions of the schema.org terms are rather narrow. "Offer" is
> > defined as:
> >
> > "An offer to sell an item—for example, an offer to sell a product, the DVD
> > of a movie, or tickets to an event."
> >
> > This seems to eliminate offers to lend, to lease, or to barter, or to
> > provide other services.
> >
> > Our question to this group is: Is the community open to proposals regarding
> > changes to definitions for terms already in schema.org? Do we anticipate
> > that such changes would/would not be disruptive? Or is it preferred that we
> > create new terms with new definitions?
> 
> It is natural for similar ideas to crop up in different areas of schema.org.
> 
> Adding new property/type associations should be relatively
> straightforward, though I'm wary of very general notions like 'partOf'
> being stretched to apply everywhere.
> 
> In the case of Good Relations, it sounds from Martin as if your usage
> is a good fit, and if the independent Good Relations spec is updated
> we'd very likely track changes into schema.org.
> 
> Just for contrast: there is another example of where Good Relations
> and the BibExtend work also overlap in their interests: FRBR-like
> models.
> 
> * The bibliographically-minded world are fascinated by the challenge
> of describing mass-produced items that carry intellectual and creative
> content.
> * The e-commerce-minded world is equally fascinated by the challenge
> of describing mass-produced items that can be bought and sold.

Two things here:

1. I believe that many or most of the "bibliographically-minded" people
working in the BibExtend group are more aligned with the Good Relations
goal of connecting the items that they're describing to people who can
use them through structured data. Connecting people to resources, albeit
with different motivations, is the common goal of librarians,
archivists, researchers, publishers, book vendors and others. Thus our
proposal to promote the citation property up to the CreativeWork
level[1][2], for example.

That's also why I was recently working on modelling "holdings" (the
available copies of a given item) as Offers[3]. We recognize the
overlaop with Good Relations and  want schema.org processors to be able
to serve up "Offers" whether those offers are copies of a hardcover book
sold via a massive online vendor, a downloadable PDF from the Internet
Archive, loans from your local public library, or copies of articles
that are only available from a licensed database.

Using, for example, Offer.sku/callnumber seemed like a reasonably good
match for a (in this case) library-specific term that would also
represent well when treated by a slightly older version of a schema.org
processor. However, given the feeble state of the "/" extension
mechanism, we will probably step back from that approach and move
forward with either proposals to change schema.org core, or publish an
extended vocabulary and use RDFa Lite to mix in the properties[4]. Or
perhaps both...

If there was more clarity on how to _properly_ extend schema.org, such
that schema.org processors that don't recognize extensions could at
least recognize the base classes and properties, then I think we would
benefit greatly (and it would potentially enable working groups like
ours to move much faster).

2. I'm not sure that trying to jam FRBR-like models into schema.org
currently "enables search engines to understand the information on web
pages and provide richer search results in order to make it easier for
users to find relevant information on the web"[5]. While theoretically
fascinating, I believe the bibliographic world has a lot of open
questions to resolve around implementing FRBR in practice before we
should attempt to impose that on in-production domains.

Long, long story short: keeping the goals of schema.org in mind should
enable us to focus on the immediately practical use cases, iterate
quickly (to adjust definitions of terms such as "offer" to be more
inclusive of non-commercial transactions, for example), and refactor
when necessary (such as promoting "citation" to CreativeWork)... much
like an agile project!

References
1. http://www.w3.org/wiki/WebSchemas/CitationPromotion
2. http://lists.w3.org/Archives/Public/public-vocabs/2013Mar/0036.html
3. http://lists.w3.org/Archives/Public/public-schemabibex/2013Jul/0083.html
4. http://lists.w3.org/Archives/Public/public-schemabibex/2013Jul/0088.html
5. http://schema.org/

Received on Tuesday, 23 July 2013 14:37:27 UTC