Rebecca S. Guenther wrote:
> I might make a METS document for what is a set of files, called a
> representation in PREMIS, also in my case equivalent to an intellectual
> entity. I may want to include some PREMIS metadata about the Object that
> is the representation. If I use Object.xsd objectIdentifier is a mandatory
> element. But my objectIdentifier is already part of my METS document as
> objID, so I have already satisfied the mandatory requirement, but I'm
> forced to repeat it in the PREMIS Object metadata. If I also want to
> include Event in digiProv to give information about the fact that I
> digitized this object, I have to include an eventIdentifier, even
> though I don't really have a system that generates event identifiers so
> I'll just be making it up to satisfy the schema.
I haven't worked with METS, but it would seem to me that if you have, or
can infer in a systematic way, unique ObjectIDs and EventIDs, then your
system is in that respect PREMIS compliant, regardless whether the
underlying expression of that is a single "record" in PREMIS XML, or
some hybrid of many schemas. If XSLT, XQuery or even SQL can construct a
valid PREMIS record from various resources in your system, there's
nothing wrong with that, surely?
If the ObjectID had to be restated, I would merely see it as a "foreign
key" in the PREMIS record that links it to the METS record. Giving
Events a (mandatory) unique ID seems to me one of PREMIS's great ideas,
so I certainly wouldn't want to see that watered-down.
Perhaps PIG needs an XML Schema/namespace/XSLT guru to propose some
elegant ways of interweaving PREMIS and other schemas (and maybe at the
same time look into freeing up some of those typedefs I've referred to).
I'd be very disappointed if PREMIS became like Dublin Core, such that an
object with no metadata at all was PREMIS-compliant!
\ Richard M Davis
/ Digital Archives Specialist
\ University of London Computer Centre
/ Tel: +44 (0) 20 7692 1350
\ mailto: [log in to unmask]