 Okay, we're going to have our next presentation from two presenters of Ralph Hodgson and Dean Alamang. Ralph is one of the founders of Top Quadrant, tools that you may know, Top Raid Maestro Ralph. And Dean Alamang is a major, major contributor to the FIBO effort. He works as one of our staff ontologists to help develop FIBO, and as you probably know, Dean has written seminal work on ontology, semantic web for the working ontologist. They're gonna talk about how to extend FIBO and they're gonna demonstrate and show that to us via what they can do with Top Quadrant. Thank you. What we will present is something that will be in the manner of a play in some ways and that I will turn to Dean at times and say, FIBO has that interest, does it? Or it has that difficulty? I present myself both as a tool vendor and an ontologist in the field, busy with the banks and busy with central banks as well as the banks, yeah? And I will be reporting on how our platforms and tools and approach to doing things like lineage and doing things like helping researchers in the macroeconomics field find things. I shall report on that. So there are two things to be covered in this presentation but the gesture we will be making is one of confirming where the boundary of FIBO is and how what I'm doing is testing that boundary in some ways, yeah? So on my part, I'm going to be quite literally wearing my FIBO hat in this talk. So I'm representing FIBO in this little drama that we're doing and so the world that I see FIBO in is that there's gonna be a lot of folks out there like Ralph, these are either gonna be vendors, independent consultants, people working inside of banks or in institutions. They're gonna be doing interesting projects in the financial world, possibly using semantic technology quite intensely like Ralph is doing and that's why this one's interesting or perhaps not so much so. I've had a lot of interesting breakfast talks since I've come here, talking about various technologies that are kind of semantic but not quite but are nevertheless interesting and FIBO should be able to have an impact on them. So given that that's gonna be happening out there in the wild, how does FIBO interact with these things and on the one hand provide value to people like Ralph and on the other hand get value back. And so in the little drama we're going to do, if this all works out well, if we may improvise some stuff, it is me and Ralph after all, there's actually a nice little feedback loop. I call it a Gestalt. Yeah, it's a Gestalt. It's like when you did object modeling, you want to make an instance diagram to test that your classes are right. When you build an ontology, you want to see how it can be put to work and that Gestalt is what we'll try to articulate a little bit. I was going to demonstrate things and there is FIBO running here on top rate edge. Top rate edge is our enterprise data governance product, all model driven. That's introduced here on this slide and I was going to demonstrate the search over macroeconomics but that's- But you've got a boost Ralph, people can come to your booth and see it involved. And I recommend it, he showed up first today. So we introduce ourselves very quickly, Dean, the book. He also is a colleague from way back in the late, actually 2002, we've known each other from that time. And top quadrant, just to say one thing about it beyond the products is we're involved with technologies that extend Sparkle, make Sparkle into a rules language, a constraint language into, and the thing that is I think very key for FIBO, when it's certainly when you get involved with blockchain we have to do some kind of interoperability validation. The working group shapes, W3C, and the work we're doing on Shackle is going to be quite significant. And I've integrated a lot of that technology through Spin and Shackle into a lot of the infrastructure of FIBO and in some of the connections here we're going to be showing that off. Both Ralph and I are pretty excited about that technology and where it's going. Shackle is saying if you're going to have this property, this value, you're going to have to have this other property. So it's defining the payload that's going on between things. That's one of its use cases. Okay. What have you been up to? Two challenges that I wanted to address. The second one, we're going to run out of time. So the first one is about how do we use semantic technology to improve search over documents, in particular financial and macroeconomic documents. And you're going to see how we've taken something like 2,500 documents from the Federal Reserve Board, from the European Central Bank, from the Bank of England, and we've used auto classification and machine learning to do that. Stop press, Ralph. Just this morning I had direct, this was somebody, from the New York Fed who was saying that she has exactly this problem. And so she's on her way to your booth this afternoon. Make sure you're ready for her. I told her to come into this talk and she is in the audience and she has the same problem that your European Fed has and so I think she'll be very excited to see what you've got. Great. Thanks for that. And the second use case is, how do we connect lineage models of the kind that banks have to do for their stress reports to specific assets, to trading situations, to reporting needs that are spelled out in the compliance documents. So because I'm not going to be able to speak to that too much, I've included the slides and they'll get into, how do you take a document and ontologize it and then discover what kind of assets it's dealing with and as I say, we're going to run out of time so we're not going to get into that second one too much. So let's focus on the first one and here we make the point that this world that you're in with FIBO is so rich in terminology that you have to kind of work that just as rigorously as the ontologies but where does the terminology come from and what does it mean to work that rigorously and it comes down to things like alternative labels, alternative terms and being able to distinguish things in hierarchies as well as in an ontological sense. So we'll see more of that as we get into this. The second bullet expresses that objective and the third bullet talks about how we want to use ontologies to go beyond just taxonomical relationships into other distinct meanings and see how that improves the search. We didn't want to start from nowhere so we went and looked for where do we find scoffs for cabarets of course barons and places like that but you'll see on a slide that follows how we were able to go from those glossaries into taxonomies but how well do they serve the need is something that we had to work with our technology with the auto classification that we do in Edge we were able to train a number of documents to see the benefit of the alternative terms and that's something we'll be touching on later. We know, Dean, when we talked that we kind of had some common ground here on those for cabarets and we'll see how that, we get a chance to talk about that later. A lot of transformation was involved in taking these for cabarets into the models we needed. There's over 7,000 terms now in our MECO vocabulary, MECO, M-A-E-C-O, yeah. So when we looked for these things and this is where Dean and I will interact a little bit more to get clear on why these things are positioned the way they are on one axis at the bottom there it's saying how ready this is in terms of being an RDF or scoffs representation on the vertical axis it's making a depiction of the terminology quality and richness. In fact, I wanted to take the word quality out, Dean. Yes, you did. I think we're one version back on these slides. Yes, these slides are one version back. In the version that's on the web we dropped the quality word and just how this is terminology. And actually it's kind of fun to talk about this slide in terms of what changes we made to it. For instance, my FIBO teammates here are gonna wonder why FIBO is sitting so low on the quality spectrum, which I did as well. Now, when we actually outlined that spectrum in terms of richness, where by richness we need to mean things like lots of alternative terms things like multi-language things and so on that's not the effort that FIBO has been so involved in. However, when we think about things like readiness in terms of have we actually poured over every explanation and made sure the explanation relates to the actual concept? Have we actually cross-referenced every definition back to Barons, to Investopedia? Have we kept track of that links that you can double-check our work? Have we made sure that we punctuated it all in a consistent way? So when you look at different definitions they all are playing the same role in your understanding. That's where FIBO has done a lot of work on this chart. And so in fact in the one that we adjusted FIBO is very far to the right because this curation has been very high and it's about halfway up towards the richness and that we have not attempted to do multi-lingual. We've only brought in a few jurisdictions that are of high priority to us. We haven't done the things that ZDW has done where it's gone across an awful lot of different fields and cross-referenced different terminology. Yeah, ZDW, isn't it right to say it's rich in its alternative labels? It's certainly very rich in its taxonomical treatment. And in multilingual. And multilingual, yeah. So the thing to know is that you should have the presentation version 10 that's on the web where FIBO appears much higher up. In fact, we dropped the top quadrant thing because that wasn't really the point of showing the slide. I hope my build show. Yeah, I hope so. If not, we might need to switch to your machine. Well, we could switch easily, I think. Yeah, we've only got 19 minutes, let's try with this. So let's talk about the goal a little bit because I think this is such a huge effort, FIBO, that it can't be damned to just having a group of people that meet from the banks. There has to be some inclusivity. I've got 7,000 terms to give you, yes? How does that work? How do I give you my terms? In what form do I make that submission? What is the process for that? That actually is the conflict on the next slide. It says we're running out of time. We're gonna move into that. So on this slide here, we make the distinction between the terminology space and the ontology space. But to make that distinction isn't about focusing on how we build model. It's what we want to do with them, how we want to put them to work in the top of the picture. There are things that these, there are use cases. And the use cases are summarized in terms of these verbs. We want models in order to discover things, to give advice, to transform things, to integrate. And the last word there is to mediate, but it actually is to exchange. The slide is out of date, so we have to tell you to look at the more recent slide. So now in the FIBO world, we've experienced the same tension and that the FIBO activity is, for the most part, over there on the right-hand side as you're facing that. In the ontology space, FIBO is written in LDL. We do a lot of trouble to make sure that the logic is consistent and expressive. However, when we put FIBO out into the world, we find a lot of people are saying, I want a machine readable data dictionary. I don't have these smarts inside of my organization to understand all this DL, or maybe a few of us do, but we don't get the penetration there. So there's an effort that Dennis has mentioned in a few of his talks, if you've been through all the FIBO ones, the FIBO vocabulary effort, where we're actually building things in the terminology space that correspond to what we've been doing in the ontology space. But now we get into the issue that Ralph was just talking about. He's not the first one who has come to us to say, I've got hundreds or even thousands of terms that I would like to see reflected in FIBO. Many of these people are living entirely on the left-hand side of this diagram. In fact, from their point of view, they wonder why we're bothering on the right-hand side of the diagram. They want to say, I've got some terms. Why don't you just pop them into the terminology space? And of course, we say, well, actually we want, did you see Sherry's talk yesterday? How many people here saw Sherry's talk yesterday? She's looking way beyond terminology. She's going to need something a lot richer than terminology to support that vision or the vision that David had this morning with blockchain as well. You're going to need something a lot richer than that. So while, yes, we want to be responsive to people on the left-hand side. We want them to participate. We want them to donate back. We're looking at sort of a longer game that Sherry and David have been talking about and that's going to be on the right-hand side. So one of the challenges the council has and those of us who wear FIBO hats is how do we keep this stuff in sync? And it's quite a challenge. Ralph is facing it in his project and FIBO is facing it in your role. OK. We could talk at length about this. But let's get to our demo. We would need to be sure about demo, I think, first. So there is a progression here from things that live as glossaries into things that live as scoffs for cabarets. And I made the point earlier about having 7,000 terms in MECO to organize and to relate to FIBO. And we will see some of that, hopefully, in the slides that follow. This is an example for swaps. Swaps get quite busy. And the one we're featuring here is a cross-currency interest rate swap. What you see on the left is a taxonomical representation of that with the details showing in the right pane. Those terms, depending on the context, you're using those terms at the moment to guide some machine learning things that will help you classify documents. The documents you're classifying are reports about those instruments. Is that right? OK. All right. So move forward to there we go. You see how this picture, just as we flash pie it, you've got FIBO up there on the center axis, yeah? And we arrive here. All right, so take the build back one now, so we get back into context. All right, so where we are at this point is Ralph's index that he's using to classify these documents that are basically reports about certain kinds of instruments. Now, we have a look at this. He is classifying these swaps according to the terminology that he has found in his documents. But we notice the one we're looking at now, cross-currency interest rate swap. Now, if you do the build, we're now looking over into FIBO. So we've switched into the ontology space. Now, we're in the ontology space. This is in some of the derivatives sort of bleeding edge FIBO that's being used in a Pufl concept just before about the State Street Pufl concept having to do with the structure of the swap. So the ontology is talking about what are the pieces of a swap and how do they relate to each other so that you could recognize them. So here we have a thing called a cross-currency interest rate swap contract, which is very similar to the term that Ralph had over in his terminology space. Now, if we move forward from here, Ralph, FIBO is describing it in structural terms. So it's going to talk about the two legs, the features in the legs, and how they relate together. And I'll zip to the next slide. Just give you a feeling for this. This is a particular swap up there at the top. It's got two legs. One leg is a floating. On the left is the floating one. On the right is the fixed one. If the figure were to get busier, you'd find out that they're both in the same currency. This is a swap with two legs. One is fixed. One is float. They're both in the same currency. What do you call such a critter? Well, we actually have a name for that in FIBO. That is actually, and if you do the next build, that is actually a fixed float, single currency interest rate swap contract. It's a mouthful, but there's nothing in there that you don't know already know. It's getting as bad as XPRL. Yeah. Now, what we've done in FIBO, here I'm showing it in the open source world. We were using SPIN with some of the rules in FIBO to take this data and draw that conclusion. In the state GPOC, you saw other technology doing this. The idea FIBO, of course, it's as technology neutral as makes sense to be. And the graphics here are top braid from top braid. And the graphics here are top braid. SPIN engine running in top braid. Here's the key point. If we look down at the bottom here, fixed float, single currency, interest rate swap contract. Ralph, do you have that in your vocabulary? Well, I have it as cross currency. I don't have the fact that it's fixed and floating. So I need to put that back now. So here is the feedback. So Ralph has some terminology. If I wanted to do research on these things, I could use his technology to find things at least all the way down to, well, what was it, the fixed interest rate swaps. But now our structural POC. Cross currency swap. Cross currency swap. Our POC says, actually, it's a more detailed thing. Now we give this back to Ralph. And what would you do with it, Ralph? Well, I'll have it organized taxonomically. But I will retain a reference to where it came from in FIBO. How do we do that? We have an ontology that lets us make cross ontology references. Vame it's called. But there are other things we can do in the SCOS world, of course, to say related. But related is a very coarse kind of property. We'd much rather have something more ontologically complete about that relationship. So essentially, it goes back, but we have to probably have to train our auto-classifier again to get more results. And then we can run that over the reports again. So we have that depicted a little bit here, except FIBO is not. I don't know the process by which I'm going to get a recovery from FIBO. That's something that I need to learn. Did I put FIBO in here anywhere? No. It belongs in this picture, for sure. We have Deidolf, and we have the Bank of England stuff, and DDW there. So here was the demo that I don't have much time to show. You see three sources of data, the Federal Reserve Bank. We're working with one of these companies up there. We couldn't say which. So you have to guess prizes for which one will be given out later. But European Central Bank and the Bank of England are there. What's going on here is a faceted search where we can show results and click on and show the document. If I were projecting on my machine, we could show this. This was done by having a training set. There you see the training set. The research papers came in through RDFization based on TECA. TECA is an Apache tool. This made its progression through our pipeline into our auto-classifier, which is based on machine learning techniques. And then we have to feed that with a terminology. He's an example of the terminology. It should be a new term. This is where it would come in at this point. Yes, it would come in at this point. And then we can auto-classify these things. So you see terminology sources coming out there. And we see European Central Bank coming in. There's another source there as well. So all this is happening with top-rate EVN, but also top-rate Edge, and is involved in the pipeline of this and integrating the whole or orchestrating the whole thing. What we see in this next slide, Dean, is just how rich these things become. This is a faceted search where you see terms on the left and terms on the right. Facets are actually coming from term-class properties. It's possible to construct ontological distinctions here that allow the facets to show up properly. All kinds of things happen in the macroeconomics world that extend this into forecasting models like DSGE and sticky price models. And Bayesian models of various kinds. But there's also the need to do the simpler things of GNP and interest rate. In interest of time, I want to answer the question you asked earlier. And by having you take me to this slide. I didn't mind as well, Dean. So how are we going to get the terminology back to you from FIBO? And this is part of the reason why we are doing a terminology effort inside of FIBO. And this is a very simple diagram. We have FIBO, the one that my hat is about. That is an ontology in OWL. It's passed all the DL tests. We're very rigorous about that. Then we have the FIBO vocabulary effort, which is actually being implemented in SPIN. I like to plug Holder's technology whenever I can that then automatically generates a thing that I photoshopped just yesterday to be called FIBO-V in this little logo down here in the lower right. It's a rather clunky logo. Autistic license, is it real? That's the clunky one that I just put together. FIBO-V is in the vocabulary from FIBO is being automatically generated. So when the State Street POC settles and the derivatives group brings derivatives back into FIBO as a published version that will come through to FIBO-V, which is distributed to vendors like you at the moment on our wiki. But we're putting together some more automated things for doing that. But that process is available, and that's why we make it available so we can feed it back to vendors like you to increase the value of your offerings. Yeah, well, it raises questions about validation process, accreditation. There are all kinds of things that happen in standardization. That's right. FIBO will have to attend to it. That's what we attend to. And that's why FIBO-V is coming out in a very controlled process. So what we end with is that FIBO can be a vocabulary, just like we have ZDW. It can live like this. Yes, and so here we're showing off FIBO-V. As a vocabulary, these are concepts that are related to one another in a broader, narrower structure. And so all of these terms are terms that have come from FIBO and have gone through the FIBO rigor, to get back to your point about rigor. But they do have quite a bit of structure to them. We're down to our last minute. With one minute, I will show one thing about lineage, how to take contractual documents into an ontological form so that the lineage model of the kind you see there, working with Oracle and Big Data and all these technologies that the banks have to be busy with to get a compliance report. I'm looking to have FIBO provide the richness of obligations, permissions, prohibitions. Many more properties start to happen when you get into a compliance ontology. And I leave you with that future thought. More than just the taxonomical world, we have something to contribute on the ontological world. So with that, I shall end. And we have 40 seconds for questions. Questions. The question is, did we use any of the ontological structures to guide the auto-classifier? The answer is yes. We feed the Maui with customization that allows us to not only have the SCOS ontological constructs, but also our own ontologies that can guide or give more precision to relationships, basically. Because instead, if SCOS related, we were able to say, identify as asset, yes? Or terms like that. Properties where we can make the asset connection. As a type connection, yes? Second question. Mike. Is the question, can FIBO have the RICO ontology to consider for foundation work of FIBO? I don't see any issue with that. I would want to discuss this a little bit with Dean. When you search on the web for regulatory compliance ontology, there has been some effort there. And I would like to talk to the people that did that work to check what we've done, yes? And have that as part of it. That's the sort of effort that we have. I don't know the mechanism on by which I would do that. But I think we'll hear this later today of the governance process. Thank you very much. Let's give a round of applause to Dolph and Dean. Thank you.