Dear all,

At the Portuguese National Archives we are developing a digital repository for preserving authentic digital objects based on Fedora Commons. The repository fully supports PREMIS and has adopted all of its current terminology.  For example, we completely abandoned the term "digital object" (because it was dubious) to begin talking about Representations, which are composed of Files, which are composed of Bitstreams/Bytestreams.

I'm not sure which terminology other institutions are using both locally and internationally, but to us at the University of Minho and the National Archives, the PREMIS terminology as it is now makes perfect sense.

Best regards,
Miguel Ferreira

- -
Miguel Ferreira (researcher)
Department of Information Systems
University of Minho Campus de Azurém 
4800-058 Guimarães, Portugal
http://www.dsi.uminho.pt/~ferreira/
Phone: +351 253 510 261 / FAX: +351 253 510 300 / Ext.: 510261

On Oct 28, 2007, at 3:27 PM, Evan Owens wrote:

Hi Priscilla,

 

We use rendition rather than representation as the term in our own work at Portico.   I don't think that the GPO's use of rendition is problematic:  a PDF rendition is a single format single object while an HTML rendition is a compound object made up of one or more files.   In our internal content model, the logical object (a journal article) includes a basket of components (XML, PDF, GIF, TIFF, JPEG, MPEG, etc. etc.) which can make up multiple renditions (print, web HTML, web PDF).  We classify all the pieces in this basket by function:  PDF are page renditions, HTML are web renditions, and most of the rest are components referenced by the renditions.  We are just now encountering for the first time PDF files with links to external objects such spreadsheets; it looks like the PDF is getting to be more like the HTML in terms of compoundness.  Our basic assumption for e-journals is that there are multiple renditions of a given intellectual object; that isn’t true for all kinds of content of course.  But I still like rendition as a term.

 

I recently found a diagram from November 2003 that I did for one of the early PreMIS discussions trying to sort out migration relationships (horizontal) and compound rendition relationships (vertical).  We didn’t end up implementing this exactly this way at Portico, but the picture is interesting.  I’ve included it below.  

 

---- Evan  

 

Figure 1

<image001.jpg>

 

 

 

-----Original Message-----
From: PREMIS Implementors Group Forum [mailto:[log in to unmask]] On Behalf Of Priscilla Caplan
Sent: Monday, October 22, 2007 1:00 PM
To: [log in to unmask]
Subject: [PIG] should we use a different term for representation?

 

This question came up at the last Editorial Committee call and we

decided to test the waters by asking the PIG:  Should we use a different

term for "representation"?

 

In PREMIS, a representation is defined as a "Digital Object

instantiating or embodying an Intellectual Entity. A representation is

the set of stored Files and structural metadata needed to provide a

complete and reasonable rendition of the Intellectual Entity."  All

semantic units in the PREMIS data dictionary are defined as appropriate

to Files, Bitstreams or Representations.

 

The original working group used the term "representation" partly to

avoid confusion with the FRBR term "manifestation".  The two concepts

are close enough to cause confusion but not identical.  We were aware of

the meaning of "representation information" in OAIS, but we thought the

OAIS concept and the PREMIS concept were different enough that there

would be no confusion.

 

I'm not sure that has been the case.  There does seem to be some

confusion over what a PREMIS representation is, but I don't know if any

of it is due to confusion with the OAIS representation.

 

In any case, it may not be wise to use the same term with two different

meanings within the same (preservation) community.  So, should we try to

come up with a different term for the PREMIS representation now, as the

Data Dictionary is being revised?  Or leave it be, since nobody has

actually complained about the term (that we know).

 

And those who think we should use another term, what term might be

preferable?  I have thought of "rendition" but the US Government

printing office uses that term in their FDsys with a different meaning,

as an "Instance of a publication expressed using a specific digital

format".  Something has to convey the idea that a representation may

consist of more than one file, and that the files can have different

formats.  Perhaps "assembly"?

 

p