Linking research & learning technologies through standards

Late in 2009 Education Service Australia (ESA) handed Link Affiliates an interesting brief: to write an executive level briefing on the semantic web approach to machine readable curriculum that they had been piloting in collaboration with the Achievement Standards Network (ASN).

Although we’d been working with ESA and ASN on machine readable curriculum for quite a while as part of our Technical Standards for Digital Education project , this brief was still quite daunting: explaining technical approaches to executives is never easy; explaining a semantic web approach to education executives even more so. We persevered, however, and in May published a paper called “Benefits of Machine Readable Curriculum” on our website. The paper describes ASN’s approach to imposing a lightweight and flexible structure over curriculum documents, and how that structure can support a bunch of simple, but high impact, use cases such as discovering content based on curriculum outcomes, and charting student progress with respect to a curriculum.

Painting a simple picture

Although the paper is probably still not quite at “executive level”, it has been receiving some very positive feedback. When we presented it to our project focus groups and at the IDEA Lab in May there was significant interest in the approach. After the presentations, some folks in the audience said they had a real “ah ha” moment about the potential of the semantic web to support education.

Perhaps part of the paper’s success was that we took a step back from the semantic web technologies underlying the Linked Data approach and recast them in terms of a few simple concepts:

• giving identifiers to learning outcomes
• associating the identifiers with descriptions of outcomes
• creating relationsships between learning outcomes and other things (e.g. learning resources or even other learning outcomes)

We tried to explain each concept separately and show how the introduction of each concept can support a broader and broader set of curriculum use cases.

We also used one of the underlying properties of the Linked Data approach to our advantage: RDF linked data descriptions can be conceptualised as graphs. This meant we could illustrate the approach with simple diagrams that we think are powerful, even without the context of the paper.

Watching our language

We have also just released an updated version of the paper in response to some feedback from Alan Bevan. In our original paper we used a metaphor of “tagging” to describe the connection between resources and learning outcomes, saying things like “Learning resources to be easily and unambiguously tagged with relevant learning outcomes.” Alan thought that this created a bias toward learning resources that were designed with the curriculum in mind and had “hard-wired” relationships between a learning resource and curriculum outcomes in their metadata. While this is a valid relationship, Alan pointed out that

the same resource may well be used in multiple additional ways by the teaching fraternity and those uses need to be also recorded and valued

and that many resources used in the classroom may never have been designed with the curriculum in mind.

So, it looks like we were a bit lax in our use of language: we had no intention of privileging just one kind of tagging. Quite the opposite: one of the advantages of the semantic web approach is that it allows anyone to make metadata claims about anything (at least anything that has a URI). Hopefully our updated paper uses better language to describe how the approach makes it easier to “connect” a learning resource and a component of the curriculum. This “connection” can mean many things and can be captured in many places. A connection might say things like:

• this learning resource was designed to support this component of the curriculum (captured when the learning resource was designed)
• someone found this learning resource has been useful in supporting this component of the curriculum (captured when a resource is used in the classroom)

Wibert’s proof of concept

Despite the positive feedback on our paper, it was still just the opinion of a couple of IT pundits about the potential of an approach: all the pictures in the world and grand RDF designs are meaningless unless the approach can be implemented. As Linus Torvalds of Linux operating system fame famously said

Talk is cheap. Show me the code.

For that reason, we were extremely excited when Wilbert Kraan from JISC / CETIS created a working proof of concept for the approach. His “How to meshup eportfolios, learning outcomes and learning resources using Linked Data, and why” blog post shows that if curriculum, portfolio and learning resource are all exposed as RDF then it is possible to create simple SPARQL queries to support use cases like:

• discovering resources based on the curriculum
• tracking student progress through the curriculum
• mapping learning pathways to the curriculum (WIlbert’s code uses plot points from the book “Robinson Crusoe” as a learning path and maps it to the Californian K-12 curriculum!)

Where to next?

As Wilbert points out in his post,

once you have all your source data in RDF, the rest is easy [...]
The bigger deal is the data: […] more curricula need be described using the ASN.

And that is precisely were we are at in Australia: Education Services Australia have been piloting this machine readable approach because they can see the potential for it to support the roll out of an Australian national curriculum and make connections between the new curriculum and learning resources. Here’s hoping they can get approval to move from pilot to full blown service.