 Thanks for the invitation to present. I've put together a series of screenshots to give me something to talk to. I actually just got back from holiday after three days of light installation. So bear with me while we go through this. So I've put together a bunch of links first where the content has come from on this. So those will be available. So I don't feel like you need to try and grab them from the screen as we go through. So they're all there collated. And what I'm going to go through is some of the background and context and the need for the vocabularies, the parameter usage vocabularies and ontology, net CDF-LD, net CDF-linked data, and linking it all together. So eReefs is a program that's been going for like 12 years now, I think. And it's a four-dimensional or a foundational part of it is a four-dimensional model of the Great Barrier Reef, hydrodynamic modeling, biogeochemical modeling. There's a lot of remote sensing work done in situ observations. So there's an honest body of research, data sets, all kinds of things there. And part of making this data available is a visualization portal that we use. And that's backed by a data-brokering layer API, which has a swagger interface for application use. And this sort of brings together a lot of data sets to make them accessible to people. So this visualization portal has a data browser in it, which uses this data-brokering software that we've developed, which regularly goes away and scans a bunch of data sets and creates a case of metadata. So where we want to get to is that these data sets are sort of open, accessible, findable, and that is like human readable. So when people are looking for things, they don't necessarily need to know strange encoding that perhaps the modelers have used in the underlying data sets. So if people are looking for chlorophyll data sets, they don't necessarily need to be searching for C-H-L-A or C-H-L-something, and they'll still be able to find those things. So when we're browsing for data sets, it yields a whole bunch of data sets from different sources. And amongst those, we've got this particular data set here, which is biogeochemical model result. And within that data set, it has a whole lot of variables. And through this interface, this sort of user interface, we get some of that metadata that's been cached about sort of labels and descriptions, temporal and spatial information, that to the side. Behind the scenes, the actual sort of structured metadata that's been cached, we can sort of go and look at that. And we've got this metadata result here. And within that, we've got temporal and information, spatial information, units, variables, and information to where the actual data set is coming from. So just gonna, yeah, talk to the sort of the underlying data set for this particular one. So Threads is a piece of software for data service for NetCDF data. I'm not sure how familiar people are with NetCDF, but it is a multidimensional structured data format that is quite common with atmospheric and oceanographic modelers. So it provides, it's built on the hierarchical data format. I think it is HDF. And it provides, yeah, multidimensional, multivariable. It has sort of headers in it, lots of metadata. And this Threads application is a way to provide web services to that, and we get this kind of web view of that underlying data set. So this particular data set has a dizzying array of 356 variables. And these are the variable names that are in the data set. So not necessarily user-friendly. So this is one of the sort of main drivers for needing vocabularies to try and make all of these variables accessible and describe things and provide extra context about, like, those variables and units and what they are in relation to one another. And to be able to leverage some hierarchies to say, like, okay, well, this variable, it's actually not just chlorophyll, but it's related to nutrients and it's within the context of the ocean and it's something that's been modelled versus something that's been observed, that sort of thing. So recently we've been moving towards using the parameter usage vocabulary from the British Oceanographic data center. And this parameter usage vocabulary has this semantic model of a property of an object in relation to a matrix by a method. It has some variations on this as well, but that's the core sort of model for it. So each of those elements come from vocabularies and then that whole thing on its own also forms a vocabulary term. And this is a detail on that model sort of showing all of the vocabularies. So everything we're seeing there with the S0, one, two, three, to 26 are all these separate vocabularies that are all being linked together with this conceptual model to derive singular terms for describing everything. So my understanding is that originally the PV vocabulary, they had the approach of saying, okay, well, we're gonna combine multiple vocabularies to create terms as required. And the actual sort of structure behind that, they were capturing this sort of broader and related terms. The PV ontology extends on that to say, well, let's keep that relationship and have a structured way of describing that. So that rather than just saying we have these broader things in terms of the matrix and the object and the parameter, let's explicitly say what those relationships are between those vocabularies within the term. So rather than just saying that we have a, sorry, we've got a E01 term, which is the total chlorophyll in the ocean calculated by a model. And underlying that it just has a broader relationship to water bodies. What we do is in the PV ontology, it explicitly says of the relationship between this term and water bodies that it's the matrix that it's within. Hopefully I'm getting that across. So to marry the vocabulary terms to our data sets, we need a structured way to describe it within the metadata of our data sets and a repeatable way that people and applications understand how to interpret and resolve things. And so that's where NetCDFLD comes in. And that is the, as it says here, an approach for using linked data and descriptions in the metadata of the NetCDF files. But what that looks like is within the NetCDF model output files, we'll have a header which provides this context for like the PUV prefix. And then within a variable, we have a PUV double underscore units of measure. And then that will resolve to describe something. And then that variable also has a term associated with it. So PUV double underscore UOM following the NetCDFLD conventions resolves to this URL. And then when an application follows that URL, it will reveal its label to say, oh, PUV double underscore UOM means scale unit of measure. So the link data allows us to sort of follow these conventions. It allows the underlying things to be extended and have descriptions on things to say, okay, well, these units of measure have a constraint on them to say that they should come from this list of vocabularies. And then you can put in things like quality assurance and quality control checking of whether things are following those standards and rules. So for this unit of measure, we've got a URI there for that one. And that results through to the Bodsee Vocabularies server. And that reveals a label of degrees East for that. And then we have all this extra context about alternative labels, whether it's a current definition or if it's been replaced by another one and other related terms to it. So this is the human readable interface. It has a bunch of different formats that are available using JSON-LD, RDF, turtle. So going down to another variable within this dataset, I'm looking at the chlorophyll one there, which is CHL underscore A sum, which is total chlorophyll that's been modeled. So we have their PUV parameter and there's a URI for that. And the process that we went through because that didn't exist within the vocabularies was we undertook a contract with Bodsee and they undertook to create terms for us. And the mechanism that we used for that was just in public on GitHub where they would create issues to discuss things. And they create an issue like this. And then we were able to just sort of sit back and allow conversation between the Bodsee team and our modellers to sort of ease out what the term was, what it was actually describing. And then the Bodsee team could actually create that term for us and then that would feed back and the modellers would use those terms within their model output files. So for that chlorophyll one, this is where we ended up with a labeled term that comes from that URI and it has some broader and related terms. So at the moment, what I was sort of flailing around trying to describe earlier, the P-U-V vocabulary understands that there's these broader and related things but the P-U-V ontology is about making it more explicit about what is that relationship rather than simply broader, it's about saying, oh, it's the water body here isn't just a broader term, it's the matrix that this parameter has been measured in and the thing that's been measured is concentration and the thing that that measurement is of is chlorophyll A. So you're getting much more explicit with that and then we can provide a sort of richer metadata interface and search interface to these items just by following through the linked information for these terms. And this will go through to the search interface as well. So I don't want to partway through implementation of it but conceptually where we'll get is if somebody searches for water body, they're gonna get results for all of these things like this concentration of chlorophyll because there's this relationship between all these terms. So by integrating vocabularies and linked data then get this sort of rich graph where we can search through things and make these things findable and accessible. Yeah, that was the search interface I was talking about. So yeah, at the moment, searching through our visualization portal for chlorophyll yields 294 results and the intent is that regardless of whether that word chlorophyll turn is actually in the label or the description of a data set. If we've got that link term from the vocabulary and that has a relationship to chlorophyll then it should be findable through here. So that was about everything I wanted to go through. Just wanted to acknowledge all the efforts of the Siro team involved and Bodsy and the fact that we're really standing on the shoulders of all the work that's previously been done on linked data and ontologies and vocabularies. And yeah, thanks very much for your time.