Developers entries help us explore new possibilities in discovery

September 15, 2011

It really was a tough call to pinpoint a clear winner for the #discodev competition. After we gave people a bit more time, using some of the August lull to work on applications, we ended up with a really good array of entries, demonstrating a wide range of possibilities. A key judging criterion (obviously) concerns the usability of the application. But judging aside, I am personally less concerned with how usable a rapidly developed application is – and some of these applications have worked very effectively with complex and often dense datasets – but how much they get me thinking about potential use cases and benefits.

To a large degree, the Discovery programme is about identifying the potential, and where appropriate finding ways to build on someone’s seed of an idea. Applications such as Yogesh Patel’s experiment with Archives Hub linked data might only scratch at the surface of the dataset but they still prompt us to think about some of the great potential that exists. Along with What’s About it hints at the potential of combining historic and contemporary geospatial data to provide new routes through to content; to explore the world of ‘exploration’ spatially as opposed through the linear and hierarchical structure of the archival description. I think the archival community especially is hungry for examples to help us get past some of our entrenched thinking about what discovery interfaces looks like. Along with initiatives such as HistoryPin, OCLCs MapFast these applications give us something tangible to react to and explore ideas around discovering library, archival, or museum data geospatially.

We’re also learning more about the potential for Linked Data. The entry from Mathieu D’Aquin, Discobro, compliments the research and development activity of the JISC-funded LOCAH project perfectly in this regard. These are projects that enable the archival community see how EAD rendered as linked data can become more embedded within the wider web of data; and instantly (it seems to me) we’re forced beyond the finding aid and document-centric mindset, and thinking about our descriptions as data that needs to be interlinkable to be found and used. It is remarkable how well Discobro works. My own search for the Stanley Kubrick archives in the Archives Hub using the bookmarklet immediately provided multiple links out to DBpedia entries on Kubrick’s life, cinematography, and films. All this is not achieved through a manual mashing of data, but an automatic ‘meshing’ that can scale (which is perhaps one of the most heady promises of Linked Data).

Will Linked Data be The Way Forward? The jury’s still out, but applications such as Discobro,  and others help us understand in much more tangible terms what benefits might be delivered.

And some applications demonstrated benefits that we can work on delivering much more immediately. For me the stand out here is the Open URL Router Recommender developed by Dimitrios Sferopoulos and Sheila Fraser at EDINA . My brain’s whirring with the possibility of how we can include this as a functionality into article search services at the local or national level (for example, embedding it into the newly designed Zetoc which will be launched later this year). The use case for recommender functions is already proven, although we have more to learn about such functions in academic and teaching contexts, but what EDINA have demonstrated is what you can achieve through the network effect – gathering data centrally. Patterns and relationships between articles emerge that are not readily available through other means. It’s simple, and the data’s already there waiting to be exploited. As a result we can provide routes through to discovery based on communities of use, disciplinary context, and not descriptive metadata alone.

Neeta Patel’s simple visualisation of the MOSAIC circulation data demonstrates something similar – through my involvement with the SALT and Copac Collections Management projects, we know that libraries are already using their circ data (if they collect it) to inform collection management decisions, but that often this work involves scrutinising spreadsheets and figures. Visual views of the data can really help support such analysis, and give that at-a-glimpse overview that can often tell a whole story.

There’s obviously a lot more that could be said about these entries (I wish I could touch on them all) and hopefully we’ll hear some views from my Discovery cohorts.  I’m now interested in seeing what conversations now open up as a result, and what practical work we can carry forward through new collaborations.