ORE Data Model

Feb-12-08, Birkbeck, London
http://www.openarchives.org/ore/

Builds on: web architecture, RDF, named graphs
Web arch: resource-oriented view of problem space
Use of URIs
Uniform interface
Exhange of info about resource state (representations)
Hypermedia – representations provided

Named graph: RDF graph is resource which can be named by URI
Must express relationship between resource map and aggregation
Relationships between aggregation and aggregated resources
Basic metadata about resource map

May express:

• Additional metadata about an individual resource

• Metadata about aggregation, relationship with other resources

• Metadata about aggregated resources and relationship with other resources

Small RDF vocabulary defined by ORE resource map
Aggregation graph as formally defined resource map

Minimal DC fields used in ORE: DC:creator / DCterms:lastmodified

Other ORE Docs

1. Resource map profile of Atom

• profile of Atom syndication format

• supports subset of resource map features

• GRDDL-able i.e. mapping to RDF/XML via XSLT

• Currently only serialisation format specified by alpha specs

• Beta specs will probably include guidelines for RDF syntaxes

2. Resource map discovery (representations of collections):

• conventions for disclosure

• OAI-PMH, sitemaps

• Links from representations of aggregated resources to resource maps etc

Current / upcoming events and activities

• ORE TC is currently preparing beta versions of specs

• USA open meeting – 3 March, Johns Hopkins University

• European Open meeting – 4 April, University of Southampton (post-OR2008)

• Mellon Foundation funding for exploratory work (currently funded projects: Michael Nelson, Old Dominion; Tim Cole, U of Illinois; JISC – tbc)
  

• ORE TC response to comments

• Version 1 specs by Sept 2008

ORE in short:

• a ‘webby’ approach to compound objects
• built on a resource-oriented perspective
• simple but extensible
• Open questions:
• Relationship with IMS, DIDL etc?
• Seed of a more resource-oriented approach to ‘repository’?

Discussion

Audience member commented: Some other approaches, e.g. METS http://www.loc.gov/standards/mets/, are not good at expressing relationships, which ORE is very good at. ORE allows you to associate arbitrary objects – it is extensible – potentially, it provides a way to combine the best of both worlds.

Sarah Currier asked: Why doesn’t ORE include sequencing? A: Interesting issue: the decision was made to produce a minimal spec (but it is extensible).

Scott Wilson added a follow-up comment: Potentially, a ‘sequence’ is a different sort of thing from a ‘map’. A resource map is itself a different sort of resource. You can’t make a hierarchy in ORE – it has to work via recursion. In IMS Content package, the manifest describes everything from top to bottom –ORE is not hierarchical. ORE identifies components via enumeration (‘this’ is a component, ‘this’ is a component, etc).