I have just returned from a very interesting workshop organised by the Knowledge Exchange organisation, on the topic of Interoperability and Institutional Repositories. There were around 70 experts from the 4 countries involved in Knowledge Exchange (UK, Denmark, The Netherlands, and Germany) discussing the following broad topics in the context of interoperability:
- e-theses
- OAI-PMH
- Research Paper Metadata
- Usage Statistics
- Exchanging Research Information
- Author Identification
The findings of each group should be made public shortly, and I will be sure to post the location of any resources that I am aware of.
In the mean time I can present only the outline of the findings of the group I was in: Exchanging Research Information. This was focussed around the possibility for integration or interoperation between Current Research Information Systems (CRIS) and Open Access Repositories (OAR). There were representitives from both communities, and a large part of the meeting was for each of us to understand the other. The Common European Research Information Format was introduced to us, in light of the upcoming release of the latest revision.
It was initially felt, especially by the CRIS community, that interactions between CRIS and OARs would be very one way, and that the CRIS would simply make available the relevant information for the OARs. This doesn't strike me as being the definition of Interoperability, and so it was necessary for us to examine what was really the relationship between the data held by each system.
The approach that was taken was that a simple use case was analysed for the following features:
1) What information it would need to encompass
2) Where the information could be obtained
3) Where the information would be of interest
The use case is the traditional repository use case of "Deposit", although it was necessary to formulate this in a more general way as a "Publication Registration Process". This allowed us to successfully abstract away from where the User Interface for such a registration process lay, and thus to take away some of the arguments over whether this was the domain of the CRIS or the OAR.
Throughout the meeting, the discussion was very wide ranging, but out of it were extracted some important similarities and differences between CRIS and OARs. The most basic formulation of the key difference is as follows:
The CRIS's primary interest is in high-quality, accurate metadata, while the OAR's primary interest is in content, and can live with a lower-quality metadata. This exposes two things: how the CRIS can be of benefit to the OAR, and how the problem domains do not overlap quite as much as it might first appear. My conclusion from this is that the interoperability we are talking about is actually about finding the layer at which these two systems domains can be stitched together for the benefit of the research community.
With this discussion under our belts, then, we enumerated first the information that CRIS are interested in, and then the information that OARs are interested in. The following list is not exhaustive, but gives an example of the differing perspectives:
- CRIS
- project information
- bibliographic metadata
- researcher role
- scientific impact
- OAR
- bibliographic metadata
- administrative metadata (technical, preservation, etc)
- collection/group information
- full-text / content
- {persistent} identifier
The resulting analysis of the use case showed that information needed to come from all corners to achieve this process, including the special case of author information, which may come to the process from yet another system, albeit via the CRIS.
The general consensus of the meeting is that a working group needs to look closely at the interactions going on in this and other use cases, and specify some set of interfaces and content models that can allow for interchange of the relevant data. This should be followed by a reference implementation and service. It was proposed that the basis for a project looking at these issues might consider e-theses and other grey literature, as they may prove to be the easiest place to start.
It was good to see plenty of crossover between this and other strands. The bibliographic metadata obviously mattered to the Research Paper Metadata group, while starting with e-theses and grey literature will matter to the E-Theses group. That author names may have to come from some third-party system may well be connected to the Author Identifier group, and since interoperability is of essence, you can barely go any distance before considering at least the base problems which OAI-PMH addresses.
All in, an interesting meeting, and I'm looking forward to seeing the reports that will be published by the group moderators in due time.
