MPEG21 Digital Item Declaration ISOIEC 210002: an overview

1
A view from the trenches LANL Repository
Jeroen Bekaert and Herbert Van de Sompel Research
Library, Prototyping Team Los Alamos National
Laboratory
2
Introduction

• MPEG-21 DIDL documents used as AIPs in LANL
  Repository
• Worked on XSLT to map from LANL DIDL to METS
• Intensive import of third party data
• Not yet had the need to import METS IMS from
  other repositories
• Typically, per information provider
• different mechanism to transfer content
• different format to represent content
• gt data transfer, conversion ingestion are
  major processes
• Exploring interoperable Archive Export/Ingest of
  structured content from other repositories
• transfer protocol OAI-PMH
• transfer format profile of MPEG-21 DID

3
Interoperable Export/Ingest Exploration

• Projects
• Mirroring of American Physical Society collection
  at LANL
• DSpace OAI-PMH/DIDL plug-in
• Fedora DIDL plug-in
• mod_oai Apache Web server plug-in
• Mapping of
• APS publication (APS AIP),
• DSpace object (DSpace AIP),
• Fedora object (Fedora AIP),
• web resource (Apache AIP)
• to MPEG-21 DIDL documents (OAIS DIP/SIP).
• MPEG-21 DIDL document profile used is not the
  one used inside LANL Repository application
  neutral DIDL Transfer Profile

4
Interoperable Export/Ingest Exploration approach
5
Interoperable Export/Ingest Exploration issues

• Understand
• document model of archive1
• transfer document model (DIDL Transfer Profile)
• in order to be able to map AIP1 to DIP
• Understand
• transfer document model (DIDL Transfer Profile)
• document model of archive2
• in order to be able to map DIP to AIP2
• This
• goes beyond a METS / DIDL / IMS issue
• is related to document models repository
  architectures
• is more at the level of Profiles of METS / DIDL /
  IMS than at the level of METS / DIDL / IMS per se

6
Interoperable Export/Ingest Exploration issues

• Examples
• Structural
• DSpace (item bundle bitstream) DIDL (Item
  Component Resource)
• Identifiers
• Are identifiers OAIS Content Identifiers, Package
  Identifiers?
• Are identifiers identifying metadata, bitstreams,
  the document?
• Lacking identifiers
• Creation dates
• Are creation dates those of bitstreams, of the
  Content, of the Package?
• Inconsistencies
• Digests for bitstreams, not for metadata
• What would help A description of a document
  model (archive1, Transfer, archive2) in terms of
  a reference document model

7
Interoperable Export/Ingest Exploration issues

• Manner in which to express specific types of
  information in the Transfer document model.
• Using
• self-defined XML Namespaces
• specific capabilities native to document model
  on which Transfer profile is built
• doesnt feel very interoperable.

8
Interoperable Export/Ingest Exploration issues

• What would help Share plug-in vocabularies
  across document models for Transfer
• Structure rather than relying on built-in
  hierarchical capabilities of say DIDL express
  via interoperable vocabulary to express
  relationships
• Digests rather than using METS _at_checksum, use
  W3C XML Signature
• Identifiers, Creation dates, etc use XML
  Namespaces across Transfer document models
• Rights rather than using METS rightsMD
  element, use XML Namespace(s) about which
  everyone knows its about rights
• To put it differently how much of your language
  does someone else need to understand in order to
  understand what you say

9
Interoperable Export/Ingest Exploration issues

• Granularity of an object
• e.g. object a complete web site in archive1
  Object single datastream in repo 2
• how to convey this difference in granularity when
  transferring data
• can Content Identifiers play a role?
• What to share
• which information should be conveyed?
• which should remain local to a repository?
• is this application dependent (archiving vs.
  dissemination vs. submission)?

10
Interoperable Export/Ingest Exploration concrete

• Reference model descriptions on how document
  models relate to reference model
• Interoperable
• relationship vocabulary
• digest approach
• approach to express native MIME type of
  compressed bitstreams
• approach to convey identifiers
• approach to convey creation dates
• These things might also be used in Intranet
  Profiles of document models