CETIS Semantics Panel session

November 21, 2007

Lorna Cmpbell introduced this panel session by asking that we think about the need to feedback from the session with a picture and a sentence for the final plenary. From the earlier talks there were already several candidates: yellow brick road, Emerald City, follow your nose, …

Lorna also asked us to think “What can CETIS do? ” “How to keep momentum going?”

The first question followed from the observation that what we teach might be concepts – but what we learn is not concepts. Learning cannot easily be expressed semantically. E.g. how to represent the subtlety in a book (such as The Wizard of Oz) and model it.

David answered that the semantic web is about machines and people exchanging information rather than describing everything. Simon added that e.g. what a student knows can be modelled just not completely. Diagrammatic representations help people think.

Martin described his software follows lots of metadata standards – why should he move up to RDF or Knowledge Mapping? What does it integrate with?

The answer was not entirely convincing as David said interoperability comes with common syntax anyway. So semantics is not about interoperability itself but across many systems. Allows exchange of data and of schemas – so what is the advantage in that? A potential use is to do automatic mapping of the data.

Simon: The vision of semantic web is of an interoperable web – but actually more likely to be many semantic webs (this view was endorsed by Alistair) Simon also referred to http://www.si.umich.edu/InfrastructureWorkshop/ as an explanation for the background to designing infrastructures. Alistair said that adopting RDF to make things easier. Simon referred back to “the tragedy of the commons”, semantic approach would be adopted if this actually solved a problem. Lorna felt that teachers do hold on to their craft and so it is difficult to say that a commons is the solution.

John Norman said that this was just an issue of time.

Michael – there is a bigger picture for why we want to do this. He felt that the approach (and also agents, natural language) can build real services.

John pointed we not be trying to answer how can we push a technology down people’s throats. Should be working on trying to help people. E.g. how to help describe subjects or create new research areas – needs building vocabularies and spotting links. Solve problems and use semantic coding in the background.

Tore: a bit worried about a problem based approach. The technology is good at some things – e.g split off training (XML can do that) but learning needs the communication and sense-making aspects that this technology can support.

eLearning perspective – JISC might need to fund to fund a development of tools through the teaching and learning programme. Lou MacGill suggested that the open rolling call mentioned in the plenary session may allow this but that there may lso be a need for a clearer planned role for connections with subject centre.

Semantic web appear server to server communication. Change in way to produce data – public information. Semantic web actually pushes against commercial drivers so public funding might be used.


My question on whether we should all need to worry about such things as RDF – the answer is that automation is being addressed. A follow up question was that JISC should build some use cases, a comment on this was we should also look for what is amazing about the semantic web. David felt some of the existing pie in the sky examples were not so useful but instead what is the next incremental step. One possible distinction would be teacher v learner support.


After further discussion Simon came up with the idea of viral semantics that would spread in current environments something extra will happen.

The end messages selected from the session both came from Simon: his Compendium map and the concept of viral semantics. Though I think Lorna had some reservations about how the analogy would work if CETIS were to then be the means of spreading the virus!