Re: Library data diagram

On Thu, Sep 09, 2010 at 05:53:00PM +0100, Pete Johnston wrote:
> > The idea behind DCAM/DSP/SF is to offer constructs and language
> > for expressing application profile constraints independently of
> > any specific syntax.  Designers of metadata are not typically
> > experts in the limitations of particular technologies like
> > XML Schema (e.g. librarians), and implementers may have good
> > reasons to want to implement a particular metadata design in
> > different or multiple syntaxes (e.g., XML Schema and RDFa).
> >
> > The DCAM/DSP/SF approach of course acknowledges that
> > alternative implementation syntaxes are not equally expressive
> > of the constraints that people may want to define.
> 
> Just to illustrate Tom's point, a while ago, I experimented
> with a (rather rough and ready - it was just an experiment)
> transform to map a subset of the constraints expressed in
> a Description Set Profile [1] into a set of constraints for
> the DC-DS-XML format [2], expressed as a Schematron schema. See
> 
> http://efoundations.typepad.com/efoundations/2009/09/experiments-with-dsp-and-schematron.html

Rereading Pete's blog post, I am struck by the observation:

    As an aside, it's also worth noting that a single
    description set may be matched against multiple profiles,
    depending on the context (or indeed against none: there is
    no absolute requirement that a description set matches any
    DSP at all). The same description set may be tested against
    a fairly permissive set of constraints in one context,
    and a "tighter" set of constraints in another: the same
    description set may match the former, and fail to match
    the latter. To paraphrase James Clark's comments on XML
    schema, "validity" should be treated not as a property
    of a description set but as a relationship between a
    description set and a description set profile.

To paraphrase the point in Linked Data terms, data (i.e.,
triples) stands on its own, and any given data may "match"
multiple profiles ranging from permissive to rigid.  Indeed,
since any given bit of Linked Data "speaks for itself" -- as
long as the underlying vocabularies and resource identifiers
are preserved -- there is no absolute requirement that the
data match any profile at all.  This may not seem like good
practice, but in the coming chaos of long-term preservation,
the links between data sets and the profiles to which they
match seem vulnerable to being lost anyway.  More grist for
our discussion of the role of constraints in RDA and other
contexts...

Tom

-- 
Thomas Baker <tbaker@tbaker.de>

Received on Thursday, 9 September 2010 17:21:28 UTC