(Apologies for cross-posting)
At this time, I can say that I might welcome this small change in METS amdSec, and maybe even some possible enhancements to PREMIS. What we need are best practices that show several different technical ways of implementing PREMIS.
Tobias rightfully brought up a good point about the modularization of PREMIS inside of METS, and that semantically it's a great fit to use the various sections like techMD, digiProv, etc. That's certainly an extremely valid way to implement PREMIS. However, if(!) anyone is implementing PREMIS outside of METS as a standalone format, it would seem to me that if they ever wanted to then reference that entire premis-as-root doc inside a METS or another wrapper, they might not have a clear construct to do so. The implementer would have to choose either techMD, digiProv, etc. to point to their entire PREMIS. Whereas if we had a mets:mdRef or mets:mdWrap as a first child of mets:amdSec we could include PREMIS data in without any potential tag abuse. Perhaps we could even tack on a @ROLE eq "preservation" or something of that ilk. That could serve a very useful role, and perhaps no less valid than the modularization/"semantically correct" approach.
This construct might make things more generic or dumbed down. I must admit however that from what I've seen of PREMIS, I think it stands on its own for a preservation metadata tool that semantically and functionally can stand independent of METS. The only thing it really lacks when used independently is the ability to link to or include descriptive metadata for an object. That's something I think PREMIS should support. Given this perhaps odd notion, I'm curious about some existing practices from the rest of the community.
1) Is anyone actually (or potentially) using PREMIS independent of METS (or your wrapper of choice)? If so, are you using the schema with the premis root element, or are you still using the modules-as-root outside of the premis root approach?
2) Is there any future desire to make format-specific extensibility in PREMIS, whether from the PREMIS board or from implementers? e.g., for an object that is a still image, does it make sense to put MIX metadata inside my preservation metadata instead of inside METS proper, even though historically speaking MIX metadata might have lived inside our METS before we implemented PREMIS? I (perhaps incorrectly) see a growth area in having PREMIS house format-specific technical metadata.
I seem to remember from our birds of a feather meeting at DLF Boston that someone (Rob Wolfe? -- forgive me if I'm wrong) mentioned using premis-as-root. How do you handle this in a METS (or other wrapper) environment, if so?
If we don't go with this addition, then we'd need a best practices suggestion on which amdSec child should serve as the container or pointer for premis-as-root data. techMD might be the next most generic solution.
I hope these ideas aren't too far fetched, apologies if they are.
Digital Project Coordinator
Network Development & MARC Standards Office
Library of Congress
LA308, Mail Stop 4402
101 Independence Ave. SE
Washington, DC 20540
[log in to unmask]
>>> "Rebecca S. Guenther" <[log in to unmask]> 7/18/2007 2:48 PM >>>
From the responses I've received so far on this issue, there seems to be
no groundswell of enthusiasm for the proposal to change METS. Consensus
seems to be what Tobias says below (object in techMD, event in digitProv,
etc.). So now I will proceed on trying to write up some best practices for
using PREMIS with METS (as it is now) to run by the community.
On Mon, 16 Jul 2007, Bronwyn Lee wrote:
> I also vote for option 1 and I agree with everything Tobias Steinke
> Bronwyn Lee
> Newspaper Digitisation Project
> National Library of Australia
> [log in to unmask]
> -----Original Message-----
> From: Metadata Encoding and Transmission Standard [mailto:[log in to unmask]]
> On Behalf Of Steinke, Tobias
> Sent: Thursday, 12 July 2007 4:52 PM
> To: [log in to unmask]
> Subject: Re: [METS] PREMIS and METS amdSec
> I would vote for alternative 1 (Use premis:object in techMD,
> premis:event in digiProv, premis:rights in rightsMD). We did this the
> same way with LMER (http://www.d-nb.de/standards/pdf/lmer12_e.pdf) in
> the METS-Profile "Universal Object Format"
> (http://www.loc.gov/standards/mets/profiles/00000011.html). If you put
> PREMIS in one METS section only, the main advantage of using METS is
> lost: A clear semantically structure of metadata. The different sections
> of METS are meant to be independent of concrete schemas. A clear logical
> distinction of descriptive, administrative, technical and provenance
> metadata is what it's all about.
> Your option 1 would change the proven structure in favor of one schema.
> Maybe there will be other reasons in the future to put descriptive and
> technical metadata in the same element because MODS or an other schema
> will be enhanced with technical information.
> If you want to use PREMIS and METS together, PREMIS should be used in a
> way to fit in METS as it is. But I think it would be essential to give
> an "official" recommendation on how to do this, because this would be
> the first step to an universal exchange format for digital preservation.
> Best regards,
> Tobias Steinke
> Deutsche Nationalbibliothek
> Adickesallee 1
> D-60322 Frankfurt am Main
> Telefon: +49-69-1525-1762
> Telefax: +49-69-1525-1799
> mailto:[log in to unmask]