[Moims-dai] SC222 from SC242
D or C Sawyer
Sawyer at acm.org
Tue May 29 12:07:42 UTC 2018
Hi John, et al.,
I haven’t changed my view as I believe the consensus view is counter productive. We will have to change the fundamental information model view of an AIP, given in section 2 and mandatory for an OAIS to be conformant. We will have to break out the Content Information to be the Data Object and Representation Information, and to show PDI only associated with the Data Object. This will be the mandatory Information Model to be used and thus it will clearly say that OAIS does not consider PDI to be associated with Representation Information after 15 years of saying otherwise. But clearly it is important as I’ve pointed out and you and apparently others agree. Given that OAIS is supposed to be a conceptual model covering the preservation topic to promote clear communication, this change will imply that the concept of applying PDI to Representation Information is now considered to be in error or not relevant to preservation, and clearly this is not true.
I was at a couple telecoms where this was being discussed and I know there were statements that people were having a difficult time with implementations approaches. As I’ve pointed out, if the concept is relevant there is a way to implement it. In this case, one could provide separate sets of PDI for the Data Object and for the Representation Information components, along with a mechanism to ensure the two components do not become decoupled.
While the proposal to drop Representation Information solves the discussion's concerns with what I have to call ‘rigid implementation’ approaches by removing one of the components, it will be a giant backward step for the purpose of OAIS and I believe it will diminish the creditability of OAIS. I recommend that it not be implemented and that some discussion, in section 4, could briefly address separate sets of PDI for the Data Object and for the Representation Information, as a valid conceptual view. There is no need to actually show this in Section 2 as I believe we should avoid complexity in the (high level) information modeling that requires conformance.
> On May 18, 2018, at 2:58 AM, John Garrett <garrett at his.com> wrote:
> I certainly understand what you’re saying.
> I think we all agree that all the PDI elements could be applied to any Information Object (and specifically to the Representation Information and the Content Information Object). And we all agree that it is valuable to do so and it should be encouraged.
> We had a fair deal of discussion about it, but I think we generally felt that in terms of what the OAIS definition and what OAIS required was that we should focus on the CDO rather than the Content Information.
> Peace and joy,
> From: MOIMS-DAI [mailto:moims-dai-bounces at mailman.ccsds.org] On Behalf Of D or C Sawyer
> Sent: Wednesday, April 25, 2018 10:25 AM
> To: MOIMS DAI List <moims-dai at mailman.ccsds.org>
> Subject: [Moims-dai] SC222 from SC242
> As I was taking a look at SC 242, I noticed the reference to SC 222. It caught my attention because I had some concern about all the changes to Content Information in a draft OAIS document that John provided to me just before the NIST meeting. Even though it is marked ‘resolved’, my comments on SC 222 are interspersed below:
> Section 1.7.2 Terminology
> From: Preservation Description Information (PDI): The information which is necessary for adequate preservation of the Content Information and which can be categorized as Provenance, Reference, Fixity, Context, and Access Rights Information.
> To: Preservation Description Information (PDI): The information which is necessary for adequate preservation of the Content Data Object and which can be categorized as Provenance, Reference, Fixity, Context, and Access Rights Information.
> There are 260 other locations where the term 'Content Information' are used. If this change is tentatively approved a table of changes will be generated. Many of the 260 uses are located in discussions of relationship of PDI to Content Information. In most of these cases it would be more accurate and cause less complications and misunderstandings if the the relationship of the PDI object was to the Content Data Object (CDO) rather than to Content Information.
> I think this proposed change is going in the wrong direction. PDI is shown in Figure 4-13, which is a conceptual model of an Information Package, and in Figure 4-15, a conceptual model of an AIP. They say that PDI ‘further describes’ the Content Information. While ‘further describes’ may or may not be the best way to describe the relationship, it does not imply how PDI components are actually implemented with respect to the components of the Content Information in any given Archive nor for any given type of Content Information. The purpose of OAIS and this conceptual model is to promote understanding and meaningful discussion, not to be constrained by particular implementation approaches or the lack of common ones. This allows for discussion, by actual implementers, as to how to address the application of PDI components to the actual Content Information implementation within a given Archive. It also allows auditors to ask how the Archive has implemented PDI components generally, and specifically for a given Content Information Package. OAIS has tried to cover the landscape of digital preservation and updates to OAIS have been to improve clarity and to correct errors, which would included removing incorrect concepts. It has not been about removing concepts to make implementations easier, or easier to audit, and I think it would be a big mistake for OAIS to move in this direction.
> Consider the PDI components:
> Reference Information: clearly this needs to include external references to the Content Information, not just to a digital object (CDO).
> Provenance Information: the origin of the CDO and of the Representation Information needed to turn the CDO into meaningful information are both important, as is the chain of custody of these components. There is nothing that says the implementation of Provenance for the CDO needs to be the same as for the associated Representation Information.
> Context Information: The Provenance discussion above applies here as well.
> Fixity Information: The Provenance discussion above applies here as well. It could be expected that Fixity mechanisms used on the CDO may be different than those on the Representation Information digital objects as there may be large differences in size and frequency of update. The mechanism used to link all the CDO components and the Representation Information components is a part of the Representation Information and the mechanisms used to prevent corruption of such linkage is also of interest. The use of registries, data bases, and outside organizations may well be involved. For example, the use of an outside organization that maintains the MP3 standard may be sufficient fixity for this bit of Representation Information, provided the Archive keeps sufficient track of the status of MP3 and the organization.
> Access Rights Information: Access Rights may be different for the CDO versus its associated Representation Information. Both seem important.
> Note that having PDI defined to be in relation to the CDO rather than Content Information does not mean that the same sort of things would not also apply to Representation Information,
> One can always do more than the model suggests, but it seem clear from the above that Representation Information should be included and this encourages implementers to think about what are appropriate implementations for a given circumstance.
> but it would not imply that PDI-type information objects are required to include the Representation Information. For example, Fixity Information can easily be applied to the CDO (e.g. a CRC over each (or maybe all) of the CDO files). It is more difficult to need to also have to cover the Representation Information Network with some kind of Fixity Information. To apply Fixity to the Representation Information is more difficult. Rep Info is often updated. It may refer off to documents held elsewhere while we feel the are available, but will need to be pulled into the locally held Rep Info when it becomes harder to get to. What is adequate Fixity for the Rep Info? Is it CRC on each Rep Info document? Is fixity needed for the off-site documents? Is the Rep Info different (and the Fixity changed)if a file is held locally rather than off-site?
> As per my comments above wrt Representation Information, these are all implementation issues for which there cannot be a standard answer. However the response should not be to remove, from OAIS, concerns for the integrity of Representation Information over the Long Term. There does not need to be a single Fixity mechanism for the Representation Network as a whole. One approach is the registration of all Representation Information digital components and maintenance of them in a database. Keeping secure backups of the database may be sufficient Fixity information.
> Another example, does anyone actually track any Context Information for the Rep Info other than its relation to the CDO?
> Nothing suggests that a Context Information object needs to separately address Representation Information any more than an external Reference needs to link to Representation Information vs CDO or vice-versa. They both are most likely understood in relation to the Content Information as an object. On the other hand, nothing prevents addressing them separately as well.
> Or does anyone generate effective Provenance Information for the Rep Info?
> The Representation Information came from somewhere and that may be of particular interest, or it may not. Also, over the Long Term, where Information may move to other Archives, such Provenance may be of interest. Better to have it as a point of discussion and possible implementation, than to ignore it.
> Although this is a significant change, I think it is an easily understood change and may in fact come closer to how archives actually understand and use PDI now. I think this change may make it easier to resolve some of the other SCs.
> My first thoughts on seeing this change in the actual document led me to think that the focus was shifting to preserving the CDO and not meaningful information. I’m not aware of what other SCs were a problem, but perhaps their authors we’re looking to OAIS for too much advice on an implementation or viewing OAIS too rigidly. We have given some examples of PDI for different domains and perhaps we could expand on this Table 4-1. Perhaps we could generate some scenarios in an annex involving 2 or 3 types of Content Information with PDI and address some possible implementation approaches, although I’m not sure this should be in OAIS.===
> MOIMS-DAI mailing list
> MOIMS-DAI at mailman.ccsds.org <mailto:MOIMS-DAI at mailman.ccsds.org>
> https://mailman.ccsds.org/cgi-bin/mailman/listinfo/moims-dai <https://mailman.ccsds.org/cgi-bin/mailman/listinfo/moims-dai>
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the MOIMS-DAI