A University Linked Data Diagram

Many of us will have seen the Linking Open Data cloud diagram, which shows which datasets are linked to each other.

This is great for showing the rough state of the web of linked data, and seeing how it changes over time.

As a developer though, it’s of less use. It serves as a guide to which identifiers for a subject might be good to use (e.g. lot of datasets link to DBpedia, it might be wise to refer to that rather than creating our own URIs for things), but doesn’t really say how that data is being used.

I’ve been thinking about how to best show this for the growing area of University linked data. As someone working on University linked data, I want to know what other Universities are using and how. I want to know what best practice should be for describing the same concepts.

At this stage, it seems unlikely that many institutions will have data that links to other institutions – most institutions will initially just have (and be making available as linked data I hope!) data about themselves.

It therefore seems more important to me that institutions talk about things in the same way, rather than necessarily talking about the same things.

This basically comes down to using the same ontologies to describe things – if a dozen Universities all describe their people using FOAF for example, that’s a great start. People can start building common tools/applications/aggregators based on this, without having to reinvent the wheel to deal with University X’s custom data structures.

I’m not great at visualisation, but I’ve started putting together a simple diagram to indicate some of the things that some Universities have made available as linked data, and more importantly, the ontologies they’re using:

(view full size)

As you can see, it’s pretty small/limited at the moment, and the only University data sets I know about are our own, Sheffield’s, and the Open University’s.

Additions welcome! I’ll aim to keep the diagram updated as much as is feasible – any suggestions on a better visualisation for this would also be welcome. I’m hoping once we’ve got a few more institutions on board, it’ll be a useful starting point for University linked data developers to see which vocabularies are most commonly in use.

Dave Challis
dsc@ecs.soton.ac.uk