Dear Shawn,

 

Here at Ex Libris we have developed Rosetta, a preservation system that is using METS and PREMIS as the AIP structure. Here is what we did in regard to your questions:

 

1) At which level do you structure your METS record for the SIP (and AIP, if applicable)?

 

[Yaniv Levi] We have one METS file for each Intellectual Entity (IE), SIP can be in many formats and structures, and can include one or more IE's. As part of the ingest process we convert the SIP into an AIP, that is one IE for each AIP.

 

2) What considerations or constraints influenced your decision?

 

[Yaniv Levi] We wanted the AIP to be self contained

 

3) What issues have you encountered in your chosen method?

 

[Yaniv Levi] many, if you would like let me know and I can send some information, just a few examples:

 

-          Where to put the structure map, on the IE level or on the Rep level.

 

-          How to utilize the FileGRP in the METS

 

-          Relations between the different types of AMD sections to the PREMIS elements.

 

-          Etc.

 

 

4) If modeling your SIP at the intellectual entity level, do you retain the same PREMIS in METS structure for your AIP or do you use other mechanisms for recording your PREMIS data?

 

[Yaniv Levi] The SIP can be structured in the AIP structure and then we will keep it, but if the SIP has a different structure we will convert it to an AIP METS structure.

 

5) If modeling your SIP and/or AIP at the intellectual entity level, how do you identify representations?

 

[Yaniv Levi] We are using the FileGRP for that, each file group represents a REP.

 

6) How do you handle supplementary material such as transcripts or OCR files? As separate representations or as part of the primary representation?

 

[Yaniv Levi] There are two options, one is to handle them as another representation with a specific type (TEI for example) another is to use it as service copies which are not part of the METS file but are transient REPs.

 

You are welcome to contact me if you would like more detailed information.

 

Best Regards,

 

Yaniv

 

 

Yaniv Levi, Rosetta (Digital Preservation System) Product Manager

Tel: +972-2-6499270
Mobile: +972-54-4999370
[log in to unmask]
[log in to unmask]">
www.exlibrisgroup.com

Please think of the environment before printing


From: PREMIS Implementors Group Forum [mailto:[log in to unmask]] On Behalf Of Averkamp, Shawn
Sent: Monday, January 25, 2010 7:43 PM
To: [log in to unmask]
Subject: [PIG] Representations in PREMIS/METS

 

The University of Alabama Libraries is in the early stages of building a preservation repository.  We are planning to use the LC PREMIS in METS guidelines for structuring our SIP, but we're looking for some guidance in choosing the appropriate level—the intellectual entity or representation—at which to structure the METS record, particularly in the AIP.  Our identifier and file-naming scheme is built around the intellectual entity, so we're leaning towards an intellectual entity-level METS record for the SIP, but we can also see advantages of archiving METS records at the representation level.  Since we are planning to archive only master files, not delivery, we currently have only one representation per intellectual entity in most cases.  Looking forward, however, we need to allow room for file migrations and recaptures as additional representations.

 

If you have implemented or are planning to implement PREMIS in METS, I would love to hear your thoughts on the following:

 

1) At which level do you structure your METS record for the SIP (and AIP, if applicable)?

2) What considerations or constraints influenced your decision?

3) What issues have you encountered in your chosen method?

4) If modeling your SIP at the intellectual entity level, do you retain the same PREMIS in METS structure for your AIP or do you use other mechanisms for recording your PREMIS data?

5) If modeling your SIP and/or AIP at the intellectual entity level, how do you identify representations?

6) How do you handle supplementary material such as transcripts or OCR files? As separate representations or as part of the primary representation?

 

Any advice would be greatly appreciated.

 

Thanks,

Shawn Averkamp

Metadata Librarian

University of Alabama Libraries

Tuscaloosa, AL 35487-0266

[log in to unmask]