 Okay, so I'm going to talk more from the end user of Vocabularies point of view. The last thing I would claim to be is an expert in vocabularies. And you'll probably see why as I go through the presentation. But we built during the major open data collection project, the ANS project, a data hub of Australian researcher marine and aquatic ecocultures. And it led us down a few different adventures in vocabulary implementation. Part of that was because the archive was conceived as a multidisciplinary data hub and it was themed around water related research. So when we talk about ecocultures in the title of the hub, it denotes a concept which is central to two of the data collections that we seeded the hub with, both of which dealt with the failure of science-based policy in the face of local implementations. So one of those was a project called Talking Fish, which was investigating the relationship of local stakeholders in the Murray-Darling Basin region. And after a severe drought, the Howard government had legislated the Water Act and that commoditised water. And in the Murray-Darling Basin, a controversial water management regime was put in with mixed environmental, social and economic impacts. And the Murray-Darling Basin authority had been criticised at that time for failing to engage with the communities. And this study was part of the response. So it was actually a social science study, but the participants included professional and recreational fishers, indigenous people who lived in that region and had for many thousand years. And that's an Aboriginal fish trap you can probably see there on the right of the screen. Also they interviewed scientists and ecologists and so there was a wider range of disciplinary perspectives in the data than simply more kind of social sciences or cultural history oriented ones. And then the second collection, similarly, concerned marine parks and it was a scientist here, Dr. Michele Voyet who was asking why did evidence-based policy fail in the implementation of marine parks when they were essentially based on really solid and sound science. So she was also moving to a social science approach. So I guess both of the projects pointed to the need for a kind of improved connectivity between quite different knowledge domains, science as a way of understanding the natural world but also social science and cultural history to investigate human practices and where the two interacted. And so we came to the decision to support multidisciplinary data in the hopes that we'd be able to bring rather chaotically bring science and social science together. One of the first things we looked at was about what was water and this was before we looked at the data that we had. We were looking for vocabs that would help us to index, you know, water-related terms and interestingly at that time we found there was not much going on with the harmonisation of water vocabs but we didn't know that Simon Cox had done quite a lot of work in that area so that's really great to have seen and I recommend having a look at what Siro have done in that area. So many of the vocabularies and we found were actually, you know, may as well have been written in stone tablets. They were on like image documents and things like that so we're really keen to go with something that was accessible in a machine-to-machine sense. And ultimately when we did look at the data we realised that well water was the theme of the whole archive but that it wasn't necessarily the subject because actually there were a lot of other subjects that came out more strongly of the water type, everything to the water research type things together. I think if we've included or if we do include more scientific data in the archive it's quite likely we'll reconsider whether water vocabularies will be relevant. But at this point the only scientific data really in the archive is some species data from ALA and also some terrestrial ecosystems, satellite images so we haven't needed to do that yet. So the main vocabulary I was going to focus on in my talk was subject. We were looking for vocabularies that were used really widely at collecting institutions available in machine-to-machine format because we were aiming for a reusable item-based collection platform and that supported Australian terms and had coverage of multiple disciplines at the level of an educated public readership. And that was sort of inspired by the ANS content provider's guide concept about you know aiming this towards a reader who had familiarity with the research area but not necessarily a specialist because that supports cross-disciplinary use. So the winners, our decision was to actually use both the Library of Congress subject headings and also Schools Online Thesaurus. Yay Liz! So why? Well firstly the answer of why LCSH is pretty amusingly answered in this quora item I found. Essentially because LCSH are used widely, have change management program in place and as you mentioned several other times in there, it's used widely. Again it's used widely and so that idea you know I mean interoperability is obviously a key concept in vocabularies and essentially we used LCSH for that reason. As to why Scott? Well we found it was amazing, like it's a really amazing resource. We chose it because it was there was a sparkle end point. It was set for a good audience level. It was multidisciplinary but it was sort of aiming at educated or educational users and it had an Australian flavour. And I'm showing you a Scootal screen here as well. I hope you can see it. Essentially in the Scootal screen it shows how in Scott if you do a search for Oxbow Lake, which is actually a foreign according to Australian term for a billabong, then it actually does find the items that come up that reference billabong. So it's obviously a related term or I'm not sure how it's expressed in sparkle but anyway it's obviously there is a related term in that thesaurus. Unfortunately we didn't actually get to implement the approach that Scootal has taken here in the front end or the website in our search facility on the Darmay site. I would really like to because I think it helps support a wider range of user discourses and edges is closer to a semantic search approach but we had some implementation issues around the problem of how to perform solar search but unite that with a sparkle query looking for related terms. I think I've got a clue now as to how to do that but that's not what we did at the time. So let's just have a quick look at what we did do and what we implemented. Because we were looking to be multidisciplinary in our support for different types of data, we wanted to use a link data approach and we thought that was a good way of building bridges or relations between discipline contexts. So obviously link data as we heard from Les before, it relies on URIs. So basically named entities should be linked via URIs so that helps to contextualise them and also the relationship should be described. So we were using Omeca which is an open source digital library platform and it's quite usable for technical novices and supports item relations. So on the image on the screen, you can see that it's linking, we can link named entities like people and institutions to items. But so it goes some way to providing a link data approach but certainly not far enough because it's not always possible to include that URI to identify the named entities and we needed to extend its functionality. So one thing we did was enable API connections so that we could connect to control vocabularies that returned URIs. So here's the change that we made to the Osemeca admin. We're looking at the content management system and we have an item here and we're editing the item to add a subject. And so essentially we've made it so that the lookup can actually interrogate more than one data source and return terms based on what the data librarian has typed in. So here the data librarian is beginning to type in salinity and they're getting on the top terms from the list from Scott and LCSH for that bit of text. So essentially behind the scenes what we're doing here, well the idea for this was that data librarians can check the subject scope while they go. It also means that because the URI has returned that the context of the use of that term is not ambiguous. And we used a server component called fill my list to basically it can be configured to query one or more data sources and return the URIs for the terms. So that's how we got to have more than one vocabulary which I think according to my old LIS lecturer would have been a heresy in the old days of LIS. And I don't know why that is. Was it because of technical incapacity to support more than one vocabulary for a field? There are potentially problems with it. For example, we had arguments over which was the better term, where terms were representing essentially the same concept. But I'm still kind of open to the idea of having more than one sort of recombinant lists being added to one field as a vocabulary. So I'm interested to hear what people think about that. Okay, one other thing I thought I'd really quickly run through. Sorry, I don't have, I'm running out of time. We also wanted to make fish, the fish species information something that could be semantically searched in the sense that fish can have many names. I mean, scientifically speaking, in terms of biological taxa, there's no such thing as a fish. And that should have warned us that fish species are going to be as unambiguous as we hoped, but no. We went ahead and we found that species, you know, they're obviously another named entity, so they should be amenable to link data principles. But once again, it was challenging because fish are people, for example, or the seafood industry, or indigenous people or scientists will all call the same fish species by different names. And we used ALA as our source of truth on fish species at the time. We found in the data that fish names also vary by geography. So, for example, in the example you can see on screen, we discovered that what an ecologist would refer to as a golden perch, which is the preferred common name, is also known by local fishers in the south of the Murray-Darling Basin as calyp, but known as to fishermen in the north of the Murray-Darling Basin as yellow belly. So we had to create essentially what were effectively local records so that the Dalmau Hub could maintain its sort of like a little set of world views or set of alternate names, while still interoperating with ALA. So it essentially meant a bit of a clutch, which was adding species records to our collection, but that had other benefits. So, in fact, on ALA, golden perch is Macquarie, an ambiguous, but regardless of the name you search on, your return-related items that mention any of the name variants. So that's just two examples of how we were trying to support search and also faceted search, obviously, and to some degree semantic search that would allow for different ways of different knowledge domains and different ways of talking about things. My questions that come out of this are, what do you think about the idea of like recombinant vocabularies, as I'm calling them? What do you think are the possible problems? And the second question was, I guess, about the difference between solar search and faceted search. How do we get the two to get closer together and to provide semantic search capabilities? So, thank you.