Federation of digital libraries with ImpressCMS

I was fortunate to attend a 'winter school' on Federation of Learning Repositories for Agriculture, Food and Environment in Hungary from 2-6 November. The workshop was essentially about how to make digital libraries interoperable with one another, so that people can search across collections owned by different organisations simultaneously – think 'cross-site search'. The course focused on standards and models for automating data exchange.

There's been a fair bit of talk in our community over the years about developing cross-site search, so it was very interesting to meet and be taught by people who are doing it on an industrial scale. Trainers included David Massart from the European Schoolnet project and Erik Duval, President of the Ariadne Foundation. Both are involved in establishing distributed networks of learning repositories involving universities, research institutions and other organisations. Put simply, they want to make everyone's educational materials available to everyone else, globally.

We looked at two models of federation:

• Querying: A search made on one site is forwarded as a query to a network of affiliated sites, which provide an external query interface. Each site in the network returns its own results, which are displayed in aggregate to the user.
• Harvesting: A site may periodically import the records (metadata) held by a network of affiliated sites, which expose their content via a web service. When a user runs a search on one site they have access to all the metadata in the network, but it is stored locally, so the search is much faster.

We had hands-on implementation sessions for both models, during which we set up a Simple Query Interface target, and an Open Archives Initiative Protocol for Metadata Handling (OAI-PMH) target, and also tools to run queries and harvest metadata against those targets. Something of a challenge, as we were working in Java!

Sharing data between different computer systems opens up a diabolical can of worms about format. So we spent a fair bit of time looking at metadata standards such as the Dublin Core and Learning Object Metadata. Standardising your data is difficult but it opens up whole new worlds of discovery, sharing and reuse, and maybe this is something worth thinking about for ImpressCMS at a project level.

The workshop included two demonstration sessions where participants displayed their digital library systems to others. This provided a good opportunity to wave the flag for ImpressCMS and also to get some feedback on a new 'Open Archive' module I'm developing. This started out as a 'publications' module using the Dublin Core standard metadata fields. However, it can also handle streaming audio, embeddable video, plain text records and hopefully soon images as well. The module can permit its records to be harvested via an OAI-PMH web service, so it can participate in cross-site search and distributed digital library systems.

The course was run under the auspices of the Agricultural Learning Repositories Task Force. It was organised by the ASPECT Best Practices Network, Organic.Edunet and the Food and Agriculture Organization of the United Nations. It was sponsored by the United Nations University, CGIAR, Rural Inclusion Project, iQTool Project, Organic Mednet and the European Foundation of Information Technology in Agriculture. I would like to thank them all for the time they spent with us at the course, and also for sponsoring my travel and accommodation, which was greatly appreciated.