This is Priscilla. I can answer some of these from my perpective but I
would be interested in hearing what others have to say.
Regarding #1, Premis definition of representation. You say "The DAIS
principle is 1 object = 1 AIP". I don't know what you mean by object in
this context. The PREMIS definitions of types of objects as files,
bitstreams or representations is meant to force us to be more specific
when we talk about objects. Say you were ingesting an ejournal article
that existed in two versions: 1) a single PDF file, 2) an SGML file, a
jpg image and 3 gifs. What would be in the AIP(s)?
REgarding #2, this is really interesting. You could do this in
Preservation level by "flattening" the values, so that each set of
characteristics to be preserved together would be assigned a
preservationLevel value. But we admitted that preservationLevel is
really a business rule and so ought to be out of scope for the PREMIS
data dictionary. I expect that many repositories will have much more
complicated rules than can be expressed in preservationLevel.
Regarding #3, creatingApplication. This is applicable to Files as well
as Representations, so it is fine if you only want to use it for Files.
If I remember correctly, the reason this was made applicable to
Representations is because some applications output sets of files that
together constitute a representation. For example, Dreamweaver could be
considered the creatingApplication for a web page which itself consists
of several files, each of which have their own creatingApplication.
Flash might be another example (I don't know Flash well enough to say).
Regarding #4, use of an object. The use of most objects would be to
display or play them or otherwise render them to a user. But some
objects might be meant to be processed by program, for example an
executable file (computer code) is used by executing it, a SAS file is
used by the SAS statistical program, etc.
Regarding #5, linkingEventIdentifier. This is applicable to Files so if
you want to use it only for events that pertain to file (like virus
checking, format validation, etc.) that is fine.
Regarding #6, good idea. This might be an enhancement we could make for
the next edition.
Barbara Sierman wrote:
> Hello all,
> During the last months we at the KB in Holland studied the Premis Data
> Dictionary. In our e-Depot we use metadata, but when we started with the
> e-Depot the Premis Data Dictionary did not exist. So a comparison was
> made between the Premis data elements and our e-Depot system (DIAS from
> IBM). Our DIAS system will be upgraded to the next version so it is a
> good opportunity to study the data model . Most of the Premis elements
> will be relevant for us too. However some questions were left and I
> would like your opinion on the following:
> 1. Premis definition of Representation: a set of files, including
> structural metadata, needed for a complete and reasonable rendition of
> an Intellectual Entity.
> DIAS principle is 1 object = 1 AIP. If an object has a relation with
> other objects the relevant information in the bibliographical metadata
> will create the Intellectual Entity. So Representation Information will,
> as far as we know now, always concern one object.
> 2. Preservation Level:
> At this moment in our DIAS system all objects have the same preservation
> level, we do not make a distinction. This might change in the future. As
> a result of our study into migration we have some new thoughts and I
> would like to share them with you. We think that it will become
> important to register information about the 5 main characteristics of an
> object (content, context, structure, behaviour and appearance) and to
> add the preservation level of each of these characteristics. We have not
> decided whether we would like to do this on object level (hardly
> possible) but we think in terms of "class of objects "and "collections"
> with a certain set of characteristics that should be preserved, for
> example a collection of pfds where the text is more important than the
> illustrations opposite to a collection of pdfs where the font of the
> text is of no importance but the illustrations are.
> I must confess these are thoughts and we are still discussing it.
> 3. Representation: creatingApplication
> In my opinion this element belongs to the File section and not to the
> Representation section, as it is the Application that created the file,
> despite the representation of the file.
> 4. Representation and File; "use of an object"
> In both Representation and File the expression "use of an object"
> appears frequently, sometimes in connection with modifying the object by
> the end-user. To me it is not clear what is meant by "the use of an
> object". In our current situation we do not see any other use of the
> object than an end user who is allowed to see or read the object
> (perhaps download it but that is use outside our e-Depot). As curators
> of the e-Depot we also "use" objects, for migration and emulation for
> example. That is the only use we can think of. Did we miss a point here?
> 5. Representation: linkingEventIdentifier
> According to the explanation of this element, events like format
> validation and virus checking may be registred here. In our opinion this
> is not Representation information but File information.
> 6. One practical point: I numbered all the elements (1, 1.1, 1.1.1 etc),
> which make them easier to trace in the Dictionary.
> I look forward to your reactions and will thank you very much in
> Kind regards,
> Barbara Sierman
> Digital Preservation Officer
> Koninklijke Bibliotheek
> PO Box 90407
> 2509 LK Den Haag, The Netherlands
> +31 70 3140109
> [log in to unmask]