> Date: Thu, 26 Sep 2019 07:12:45 -0400
> From: Vicky Phillips <[log in to unmask]>
> Subject: Mapping METS to PCDM
> Hi All,
> I'm looking for people who have migrated from METS to PCDM and who'd be willing to share their experiences and expertise. We're currently mapping our Fedora 3.8 METS documents to PCDM in readiness to migrate from version 3.8 Fedora to version 5 and I have a few questions around this.
> We are going to store the FOXML exported from Fedora 3.8 and are thinking of placing this as a Binary on the Parent Object and also Child Objects alongside the access copy, MODS binaries etc. Is this how others are doing this or are they placing it as a Binary on a totally separate Object (i.e. Archived Parent Object and Archived Child Object)?
We use pcdm:hasRelatedObject for this sort of thing, attaching the XML file to the related object via pcdm:hasFile. We haven't yet done this with the FOXML, but rather with, for example, a METS file describing the newspaper issue from the Chronicling America deposit package. We will likely do that same thing for the FOXML when we start migrating objects from our old repository. I think keeping the old FOXML in Fedora>=v4 is a *very* good idea.
> When mapping Physical and Logical structural maps from METS are you able to place two instances of iana:first and iana:last on an Object and differentiate between them by using rdf:ID attribute e.g. rdf:ID="PHYSICAL"? Or do you have to have two separate Parent Objects, one pointing to the Physical OrderProxies and another (with pcdm:AlternateOrder) pointing to the Logical OrderProxies?
I'm afraid I cannot help with this exact question since we have not done this. What we have done with respect to alternate orderings is to use the same newspaper page-objects in two different parent objects: (1) in the issue, and (2) on the microfilm reel where from which the issue was scanned. In that case, however, the microfilm reel is a pcdm:Object with its own first and last members (i.e. "frame in reel" vs. "page in issue"). I will say that in my personal opinion there are diminishing returns with elaborate content models in RDF, but of course it depends very much on your use-case and the requirements of your front-end applications. These "reel" objects are not really used by any of our front-end applications so they're somewhat superfluous.
> Any guidance would be very much appreciated.
> Digital Standards Manager, National Library of Wales
University of Maryland College Park