There was some good discussion, but for me, the most helpful thing was the distinction between Hard and Softy Semantics, and the most interesting was a discussion on the long term impact of Semantic Technology on Teaching and Learning in the Large.
Hard and Soft SemanticsA bugbear of mine for a while now has been the confusion that exists in the e-learning community about whether semantics is about knowledge structures for people (like concept maps) or knowledge structures for machines (like RDF/OWL documents). In the workshop it was clear that people thought about it in both ways, and it was useful to be clear about each.
Soft Semantic Technologies help people conceptualise their knowledge - for example, by using mind maps, or by tagging items in a controlled vocabulary (community tag clouds). These kind of technologies require the learner to be aware of the knowledge construction process, and to use it as a learning activity.
Hard Semantic Technologies help machines communicate - for example, by swapping RDF controlled by an agreed ontology defined in OWL. These kind of technologies might not even be visible to end users, but instead make it easier to integrate systems and develop mash-up style applications.
A standard like SKOS is interesting because it is a hard semantic description of a soft semantic activity.
Thinking of Semantic Technologies in this way makes it easier to consider the advantages and challenges for each, as hard semantics are several steps away from everyday users, while soft semantics might be included in their work or learning activities.
The Impact of Semantics on Learning in the Large
A lot of the discussion at the workshop was based around fairly short term benefits - this was deliberate, as the focus was on how to move the e-learning community forward, not speculate on the future. However, one thread did emerge that might have major consequences for education in the longer term.
Les Carr talks about the Google Test, a version of the Turing test where the question is not 'are you talking to a human or computer' but 'are you talking to an expert, or a good Googler?' The point Les is making is about the value of information in a society where information is plentiful, and therefore the worth of learning that information in the first place.
We've already seen the impact of this in education - where the web (and in particular sites like Wikipedia) have devalued the ability to memorise and recall information. It is now trivial to find information on subjects that only ten years ago would have required days in a library to unearth. This has changed the way that we assess students, and challenges the first level of Bloom's taxonomy, which is all about memorising and repeating facts.
The interesting question raised at the workshop was that if Semantic Technologies are successful, and reasoning and expert systems follow, will technology also challenge the higher levels of Blooms - such as analysis and synthesis?
At present you could ask a student to write an essay on the influence of the Royal Society and expect them to begin by finding out what the Royal Society was, when it existed and so forth - this aspect is already easy. But imagine a world in which they could gather not just facts, but an entire essay, because the system understood not just the data, but how the data was related, and not just answers, but how to structure those answers into an evidenced argument.
Such a world raises questions that go to the heart of what it is to learn, and questions the objectives and values of modern education.
I certainly don't think that outsourcing our thinking to machines is a good way to go (especially if those machines are really just glorified Prolog engines), but if there is an easy path then some students will be tempted to take it, and that is a genuine challenge for the future.
No comments:
Post a Comment