What we are trying to do today is to identify a particular type of user working in a particular type of environment looking for particular types of resources.
Jane is our hypothetical teacher profile. There followed a requirement gathering exercise:
We would like you to build a system that has these requirements, its fast, generates relevant information we are hoping for a system that allows access to rich multimedia content that’s accessible through Google, the site itself provides support, its needs to understand the curriculum data model, provides a variety of content of different shapes and sizes, it can rank and detail what other teachers found useful and can guarantee and confirm classroom readiness – being fit for purpose.
Mike Lowndes: But we don’t want to build a portal. How do we get from here to anywhere remotely like being able to have a Semantic Web structure that could deliver that, without having to build the portal?
Martin Bazely: The value of a system that we might construct is not just in finding it (learning object) but, if you harness the feedback of everyone, you are enriching those objects so they get more and more classroom ready.
Frances Lloyd-Barnes: What we are talking about is rank and users adding value by identifying the usefulness, appropriateness or relevance of the content. Its really not only about gathering the tags but signing authorship, to state ‘I’m a teacher, this is what I’m doing’ so that some one else who’s a teacher can find what the other teachers have said and go to that.
In essence it’s like My Space, as you find somebody who has recommended something and you begin to trust them because of choices that they have made so you somehow make a connection.
[Mike Lowndes stands up and starts to draw a possible semantic web search application for teachers. It begins to look a lot like what Haystack, a SIMILIE product, could be developed to do]
Mike Lowndes: [using a Semantic Web approach] a lot of the context of your query is auto-generated from [implicit in] your situation, who you are, if you’re a teacher, a female teacher, it knows dates, it knows the curriculum, because of that it [should be] more able to query more intelligently the stuff that’s out there, but right now we still need some more stuff done at the query stage, to enable the semantic web, and make it all happen.
The ‘black box’ here [on diagram], the tricky bit, is the time between launching the query and the generation of results, where at the moment we’re all thinking of the Google approach – the ‘brute force’ approach. This ‘black box’ is where the more intelligent application of querying something [takes place]. That requires the hard stuff, the multiple ontologies, the RDF etc.
What’s being adding to the brute force approach now [with Web 2.0 etc] is tagging, and feeding that back into folksonomies, and having them available, in a standardized form, in a package, is possible a next step.
We need to come back to what underpins this and that’s that our content is still in a rubbish format. We are much better than some other sectors, because we do have curators etc. doing semantics, but its still a matter of exposing it right, getting it out there in the right way so that it can be used.
Jon Pratty: What’s interesting is that we are worrying about multiple ontologies and the possibility that it might be too many to control but we can limit it if we think about what users need, they need some trusted structures.
If in the absence of building a black box we have a mid term large SW and it’s accessed through Google then you really only need one or two or three established relationships between taxonomies, for example, the major ones are the curriculum taxonomies, there are a multiplicity of relationships that bind together. Once you’ve got the major ones in place there will be some trust evolving between discoverable objects.
Jeremy Ottevanger: Plus most of what we have is might be relevant to the learning community is not going to be learning specific and we might not even bother to mark up semantically for the learning content, because we have millions of objects and photos that are very granular things. Those are not necessarily things that were going to have the resources to tag up, there’s only a small number of things that we will be able to do that for.
Ross Parry: We’ve used the scenario of learning objects and a teacher as a way of testing some of our ideas. We weren’t ever going to come up with a solution to help Jane but we’ve used Jane’s scenario to help us to start working through the problem. We came up with some interesting ideas relating to learning objects and they might still have a very important role to play and it might be learning objects or objects for learning that is a very important appropriate place for us to build a demonstrator or suggest our first pilot.
Ross Parry: Cutting though today is this issue of who tags, is it the owners? Is it the producers? Is it the users and learners? There’s been that useful distinction between learning objects and objects for learning, and as soon as you starting thinking about objects for learning all bets are off – its as though everything potentially becomes a learning object, does that mean that the whole of learning objects and specific or not within this. Who comes up with this? Who builds the model? Do we wait for education? Or do we just do it? And the final thing, do we use an existing standard? Or are we trying to build something for the museum sector?