Do you generate PREMIS:object records for the file-level as well? Perhaps you switch to a trickle-up approach where the decisions regarding the preservation intent of various bitstreams affects the containing file's preservation value and then the files' preservation values inform the preservation value of the representation. I tried to cobble together a mock-up here http://avpreserve.pastebin.com/jx1T2B3n
of a representation made up of an open format file and a proprietary file where the least committal preservation statements trickle up to the representation, although perhaps it's better to exclude preservationValue at the representation level and simply utilize preservation values of descendant objects to evaluate the status of the representation as needed. In this way modifications in preservation strategy (such as determining at some point that the library can indeed offer full preservation on file type X) would only require modification of the objects of file type X and not reassessment of all parent objects that may contain a file type X.
On Feb 8, 2011, at 11:30 AM, Beers, Shane wrote:
My organization currently employs preservationLevel on the representation level with:
1 is a unit in our admittedly questionable controlled vocabulary. Here, preservationLevel refers to the entire preserved object, and thus trickles-down to the individual files within the object. However, we are beginning to ingest packages with file types about which we cannot necessarily make the same preservation promises, and are interested in using preservationLevel for individual files within each object. However, we are questioning the best syntax to employ for this, and are interested in seeing examples of how other repositories or organizations are doing it.
Thanks for any assistance.
Digital Preservation Librarian at the University of Michigan