15:30
Lessons in cross-repository integration learned from the aDORe effort
Herbert Van de Sompel, LANL
[scholarly communication diagram from 2003 OCLC Environmental Scan showing repositories in the centre]
build value chains across repositories
a few words about aDORe
aDORe is not a product
- components of aDORe software, usable in other environments, will be released
- services build on terabytes of locally stored content (Elsevier journals etc.)
- broke tight integration between data and app
- standards-based, modular, distributed, protocol-based interactions between modules
two front ends:
- OAI federator
- OpenURL resolver (for ? compound/complex object services? not a regular resolver)
Uses hundreds of OAI repositories locally.
Basically because of the good, loosely-coupled design, you can derive insights about inter-repository compatibility.
Repositories and units of communication
- Data-oriented research: not only text but datasets, software, simulations, dynmaic knowledge presentations
- Facilitate collaboration
Think about compound objects.
- has a persistent identifier
- contains material, and metadata about those materials
- can contain other compound objects
URIs:
- minted by different repositories
- from different namespaces
- not (necessarily) locators
"I don't think it's accceptable to ask everyone in the world to use e.g. the handle system"
need XML-based representation for compound objects
- many options MPEG-21 DIDL, METS, IMS/CP, RDF...
Repository interface
- OAI interface for compound objects.
- Use OAI-PMH datastamp ~= new version.
- include provenance
This is NISO OpenURL - framework to define service-oriented applications - applications for classes of objects that you can describe with identifiers
You can pass a SERVICE TYPE in the OpenURL.
Conceptual interface is persistent.
OpenURL examples were in HTTP, but conceptually you could do this in SOAP - just pass the parameters.
Repository Registry - who is part of this federation of repositories
Object Registry - what is part of the federation
Query: get list of existing copies, and the INTERFACES to get those things
There is similar work in the area of learning object repositories.
Pathways InterDisseminator
- Dynamic Service-Oriented Overlay upon the federated architecture
Service Overlay OpenURL application - list of services that can be applied to an object (the services are decided by the LAYER, not by the repositories)
"magic engine" - a knowledge database that knows about potential properties of objects and relates those to potential services
You could have multiple service overlays in a federation.
This results in the ability to provide context-sensitive dissembinations of digital objects.
[demo. Very cool.]
If we can meet the requirements he presented, many interesting capabilities are possible.
Q: Open Knowledge Initiative (OKI) also failed.
Why not something like WebDAV, so we can talk outside the digital library domain.
A: OpenURL standard has come out of our world, but it has extreme potential.
Q: This is great work.
A: Thank you.
Comments