Yes, Les Nebane is my name from Education Services Australia and I manage the Australian education vocabularies. And I suppose as far as that seven point terms of reference goes, this is a use case of how vocabularies are applied. Quick overview. I'm going to start off by talking about the role that

vocabularies have played at Education Services as sort of a change management tool. We're going to look at the the evolution of using vocabularies in terms of using or starting to use identifies as well as textual labels and then how they evolved into something else that we're all very excited about these days.

I will look at one more sort of sophisticated linked data application, which is about something to do with curriculum subjects and information resources. And and I'll finish off with a couple of reflections on things that you know, maybe

we did well or things we'd like to do but are having a bit of trouble with. Okay, so and is the obligatory look at Education Services Australia. We are a well, we're a service provided to primary and secondary schools in particular.

We provide information resources and websites where more or less a publisher. Here's a quick look at some of our services. I'd like to say that all of these services were employing controlled vocabularies.

Unfortunately, I cannot say that but we'll we'll look a little bit of scootal towards the end of this presentation. But okay, the Australian Education vote vocabularies number about a dozen vocabularies, but the one that everyone

knows that the most well is called the schools online thesaurus or Scott and I'm going to focus on this vocabulary partly because it's a vocabulary that ESA actually has compiled itself and owns. We also host vocabularies on behalf of other organizations, but this one we built from the ground

up and also Scott is it's very feature-rich and it's employed in some fairly sophisticated applications. It's also been around for a long time. So there's an interesting story about its evolution.

I'll just draw attention to the word online in some ways. It's a bit of an unfortunate acronym we have and we're sort of stuck with it because people often think that the online refers to the kind of resources that were described with Scott like Scott was designed to describe online resources.

Well, that's not exactly what it means. All it really means is that when we publish Scott for the first time in 2004, we published it on the web and we've never printed it. That's really all it means. It's it's a source that is online we do indeed

describe some online resources back in 2004 learning objects were all the rave in school education. The term learning objects doesn't get you so much now, but they're more or less interactive content that might include

some deliberate learning design or some dedicated curriculum outcomes or might even include an assessment task or something like this. And that was all the rage 10 years ago and Scott was designed partly with describing these resources in mind.

It's difficult to find a symbol for interactivity. We happen to use a cube on one of our website. So I've used to hear but Scott is also used to describe offline resources and if you want if you look into any one of about 95% of the school libraries in Australia, you will find printed materials on shelves that have been catalogued with the schools online the source.

So that's actually where it's widest uptake is it's actually still I think print materials still make up about 70% of school libraries in Australia. More recently. Well, ESA is increasingly in the business of cataloging and publishing what

we call open education resources or OER, which is really euphemism for we're broke. We don't really have the same resources. We used to have to design interactive content.

But but this means that we do a lot with sources of Creative Commons content and particularly repurposing content that you'll find in the in the galleries libraries archives and and museums.

So I want to talk a bit about the curriculum and Scott but before I do I want to go back to say about some 2005 and just look at a basic problem that we had an Education Services Australia and that was about how to manage change in a thesaurus without without content.

And we had designed a content management system and that understood how to ingest a thesaurus in a format that we also designed. So we have a nice convenient closed system here and with the content management

system, we would publish education resource metadata that included terms from the schools online thesaurus. And so that's all working. Well, and when we added new terms to Scott that didn't break anything, but of course eventually when we got past our sort of growth period we weren't just adding new terms to Scott.

We were changing some of the terms in Scott. We will make we were updating terms and when when we did this we found that well that meant that some of our records had terms that were no longer valid. So we had a cleanup task to do I'll just take an example.

Here's a term ecology at some point in the evolution of Scott. We changed that term to ecosystems and I know that some of you will want to fight with me about this and it's there are many controversial examples. I'll just make two quick points about this in that that is as a matter of policy in Scott.

We prefer to use the names of things and processes materials Etc rather than the names of the disciplines that somehow govern them or study them. And the other point that I'll make is that in a thesaurus or perhaps

even in language generally, I don't think that there are any true equivalences there. Rather we make we make equivalent relationships within a community of practice and that and that's exactly what we did. So we changed the preferred term to ecosystems, but we retained ecology as a non preferred term.

And by doing that it meant that we could actually do our cleanup tasks a little more easier. We can do a global change for any records that contain the label ecology. We change that to ecosystem. So everything was working fine, but everything didn't work fine forever.

So what if we what if we retire or what if we change a term like ice, but we want to have to replacement terms. It's not so simple. We can't do the same cleanup task. And so in order to solve this problem, we had a look at our metadata standard again, and we realized that we were

populating our metadata with labels, but there was another subfield called identifier. The metadata standard by the way is called IEEE long or learning object metadata, and I don't recommend it

but that's just a metadata scheme context, which I think is significant and I'll talk about that again later. And so we discovered that we could be putting identifiers in well, where were we going to get? Where are we going to get identifies? It turns out we already had identifies in the software that were used to manage our thesaurus and that software was called multitests.

We don't use that anymore, but multitests was tuning out and identifier every time we created a new term. We were just ignoring it as far as we were concerned the the term was the primary data. But we started exporting these IDs with these labels because we knew that it was going to help us

do our was going to help us validate changes in our systems and also the systems used by our stakeholders. And so there was a kind of a change of thinking that came with this and we stopped talking about terms.

We started talking about labels rather than terms to emphasize that these were an attribute of something else. We were really and I think today what I actually manage is is identifies I don't manage labels anymore and manage identifies and the labels are one of the things that I have to deal with.

So those later on those identifies became you our eyes that we're all familiar with now and I suppose that the the primary use case for doing this is that anyone who's using these

concepts or these vocabularies can do so and validate their data from the web anytime they like. I'll just put that in simple terms because I think that is really the main return on investment there as is the the being able to manage change.

by distributing identifiers Okay, so that's that's a fairly simple story about what about the role of vocabularies in managing change.

And of course, we're all interested in more sophisticated applications of linked data or you are eyes and so I'll just quickly point out a few other things that we do now. What we can do is that we can we can store our preferred and non preferred labels.

It's actually easier now for us to store more preferred labels and other languages if we choose to and we have we have done that and that's a work in progress. We can even support references to media such as images which provides a concept with lots of scope and clarity course of picture tells a thousand words.

But then of course the linking to other vocabulary. So now that we're publishing you our eyes we can create machine readable links to vocabulary such as DBpedia the second one Library of Congress.

I think if you resolve that you are are you actually come up with ecology not ecosystems, but that's not a problem for me because as far as I'm concerned we're linking a concept to a concept not a label to a label. And in the third example, the Australian curriculum and this is a little different

we we don't link Scott to the Australian curriculum the Australian curriculum links to Scott. In other words, the Australian curriculum uses Scott to describe its own curriculum objectives. And this is this is very interesting use case traditionally vocabularies have been used to describe the use case.

I suppose what you might call information resources or bibliographic resources if you like and and in this scenario, we've got something different curriculum is like a sort of a framework or an organizing an organizing framework and we've used a vocabulary to describe the Australian curriculum and that opens up some really interesting possibilities.

So then here's here's that scenario where we're using something like schools on like the source to describe some information resources, but if we're

using Scott to also describe the Australian curriculum all of a sudden we can make an inference from those resources to the Australian curriculum. This relationship supports the idea of data mining. You've got a database of information resources that are tagged with Scott or even just tagged with terms that are

like Scott turns you could mine that collection and make inferences about how it might relate to the Australian curriculum. That's something that we're actually doing at Education Services Australia with our discovery portals, but it's a principle that I think can be applied elsewhere.

You've got some kind of framework whether it's a policy framework or some kind of some kind of organizing principle to the extent that you describe it with something like a subject vocabulary. You're making it easier to align resources with that framework.

We actually are doing this in a system called scootal anyone can visit scootal and if you lose the user browse by Australian curriculum feature, that's actually the what's happening under the hood when you look at various content descriptions and there are more detailed ones and you can see on the screen.

You'll get you'll get results with resources that contain those Scott turns that match the Scott turns in the Australian curriculum. And so you anyone can have a play with that. You do need a login to look at the details of a resource, but you don't need to log in to browse results and just have a look at how that's working.

Just a bit about identifies I guess this is a kind of a kind of a lesson learned I mentioned before that we we retained the resource.

They're identifiers from our old multi-test system and we preserve them in the HTTP uris and we did that because it helps with some backwards compatibility issues. But some other approaches to identify patterns will sometimes the preferred label is used as the suffix of an identifier as in this ecology example.

It's a very to my in my opinion a very bad idea to use pref labels as identifiers because as I've already described once you change the preferred label the identifier becomes I suppose it becomes misleading also you have to choose a language.

So if you've got any interest in storing labels and other languages in the identifier becomes almost culturally insensitive. Well, something like this the second example here is a kind of a hierarchy approach

and this is a different kind of another example of including semantics in your identifier. There are traits of is a trade-off including your semantics in your identifier because you're introducing some

rigidity and so you could but we Australian curriculum does this and it's it's not a bad idea. So have identifies that people can quote in a sort of a human readable context and I for example quote Australian curriculum identifies in in tweets because they're easy to communicate in that environment.

Then, you know, the third approach is make your identify completely opaque and I think this is the really purest one and you really really forcing systems to go and resolve that that I'd that URI to find out. Why what the label is you can you can have your cake and eat it to though

we discovered that it's possible to retain a kind of a human readable or semantic identifier within a concept using this property called Scott's notation and Scott's notation is a little bit like a sub identifier.

And you can go even further and create some rules under the server that respects that identify as an alternative suffix so that it rewrites to your your main identifier if we can call it that and we started doing that at

ESA some time ago, but now actually pool party, which is the software we use does this out of the box. Now if you create notations in pool party, it will create you our eyes that alternative you our eyes that rewrite to the main you are.

I haven't really been watching the clock, but I've only got one more slide. So I hope that's okay. I suppose the the other thing that I mentioned is that our revolution in vocabularies sort of technology or standards as in publishing URIs and using RDF was we sort of went ahead of our metadata.

Standards and we're actually still stuck using metadata standards that don't really support you our eyes very well and there's been a bit of shoehorning going on. We've got systems that are based on standards like I triple-e long.

And of course, we play with traditional library metadata like the mark standard and particularly mark is interesting because you look at what the Library of Congress is doing in terms of trying to evolve the mark standard so that it does support you. Our eyes is quite a lot of work going on there, but it's been going on for a long time through difficult work.

So I suppose there's something there about if you are in the lucky situation of being able to start publishing the URIs and you don't have a metadata standard adopted already you need to look for one that really does support them properly.

Yes, it's a bit of a quagmire actually and okay. Well, that's actually all I've got.