Podcast: Play in new window | Download
Subscribe: Apple Podcasts | Spotify | Amazon Music | Android | Youtube Music | RSS
Knowledge graphs are complicated. They might entail multiple ontologies, several data sources, and any number of generated graphs and other context-establishing elements.
Eric Little came up with the idea of “super domains” to understand and manage these important but ephemeral pieces of an enterprise’s semantic knowledge architecture. A super domain might establish context that holds for just a few seconds for an ATM transaction or for several years for a clinical trial.
We talked about:
- his recent move from Accenture to Knowledge3 as Chief Data Officer
- his diverse background in the consulting and enterprise worlds
- super domains, the conceptual framework he has developed to deal with generated graphs, reification, and other knowledge graph elements
- how super domains capture context
- some examples of super domains
- how super domains can help you deal with temporary, ephemeral, and stochastic data
- how his philosophy background led to his early exploration of the concept of context graphs
- the crucial distinction that needs to be made between epistemology and ontology in semantic practice
- the role of phenomenology in his conception and practice of ontology
- his take on the current state of semantic technology practice
- how generative AI has brought the importance of metadata to the fore and highlights the need for deterministic capabilities in AI architectures
- the dearth of tooling in the semantic space and the need to deliver “semantics at the speed of AI”
Eric’s bio
Eric Little, PhD is Chief Data Officer at Knowledge3. He previously was Industry Innovation Principal Director & Head of Semantic Strategy & AI for Global Assets at Accenture. He received a dual Ph.D. in Philosophy and Cognitive Science in 2002 from the University at Buffalo, State University of New York. His Post-Doctoral Fellowship at the University at Buffalo’s Department of Industrial Engineering (2002-2004) focused on developing ontologies for multisource information fusion applications. He has worked in academia as a professor in several fields at several universities, as well as held multiple management & C-level positions in the software development industry across several different industry verticals.
Having such a diverse background spanning academia & industry over the years provides Eric with a very unique set of experiences in the software development space that cuts across numerous disciplines and business verticals. After receiving his PhD and subsequently doing his post-doc, Eric has held various academic positions including Assistant Professor of Doctoral Studies in Health Education & Health Policy and founder of The Center for Ontology & Interdisciplinary Studies at D’Youville College. During this time he also started and ran his own consulting company which landed several high-profile customers across industries, including healthcare, insurance, oil & gas, and medtech.
He is a world-recognized expert in semantic technologies, data fusion applications, data modeling, analytics, and AI. He has numerous professional publications in these areas, has been featured in industry publications, and is a well-known speaker at conferences around the globe. Before working at Accenture, Eric co-founded and was CEO of LeapAnalysis, the world’s first fully virtualized semantic search & analytics data science engine, which was named the #3 Most Innovative Data Science Company In The World by Fast Company Magazine in early 2021. He also was named Most Innovative CEO by Global CEO Magazine. He brings this knowledge and his passion for innovation everywhere he goes, developing new technologies and furthering the growth and success of his client base. Eric has a very simple goal in life – just change the world by making things no one has seen or thought about before.
Eric is married and lives on the barrier island in Indialantic FL with his wife Jodi and their 2 cane corsos, Lemmy & Eddie. His daughters Gabrielle & Claudia are recent grads from University of WI Law School & Florida State University School of Business, respectively.
For his personal life, Eric is a former semi-pro musician and is an avid guitar player & collector. He currently has a small digital studio at his house where he still writes and records songs (fun fact: his previous band, Satori, from the late 80’s-early 90’s is listed in The Encyclopaedia Metallum). He is also a motorcycle enthusiast and likes to tinker on his vintage bikes when time allows.
Connect with Eric online
Video
Here’s the video version of our conversation:
Podcast intro transcript
This is the Knowledge Graph Insights podcast, episode number 56. A knowledge graph is a complicated thing. It might entail multiple ontologies, several data sources, and any number of generated graphs and other context-establishing elements. Eric Little came up with the idea of “super domains” to understand and manage these important but ephemeral pieces of an enterprise’s semantic knowledge architecture — context that might hold for as little as a few seconds for an ATM transaction or for several years for a clinical trial.
Interview transcript
Larry:
Hi, everyone. Welcome to episode number 56 of the Knowledge Graph Insights Podcast. I am really delighted today to welcome to the show, Eric Little. Eric has most recently become the Chief Data Officer at Knowledge3. And welcome, Eric. Tell the folks a little bit more about what you’re up to and your transition into this new role.
Eric:
Yeah, thanks, Larry. It’s been a wild ride. It’s been about a month now. Yeah, so I left my position at Accenture where I was leading a lot of our strategy practice around semantics for AI. But I happened to get back in touch with an old colleague of mine, Tom Plasterer, who I’m sure you and everybody knows. So Tom and I have had a long relationship. I mean, ironically enough, it’s kind of funny, we’re both actually from Wisconsin.
Larry:
Oh.
Eric:
So we’re both Packers fans. He was born in Madison and grew up there. I was born in Green Bay and grew up there. So we come from kind of a similar Midwest background. We both root for the same football team. We both got into semantic scientologies. We’ve both worked a bunch in pharmaceuticals and life sciences, so it’s kind of strange. We always refer to each other as sort of brothers from another mother kind of thing. And so Tom approached me about the company and it was a really great conversation and a good idea.
Eric:
And it really got me thinking because it was very similar to a company I had co-founded and was CEO of before called LeapAnalysis, which was a data virtualization company where we made Fast Company’s number three most innovative data science company in the world in something like 2021, I think. But I went to Accenture. I’ve been there for about the last five years. Accenture was great, really good company to work for.
Eric:
Massive company as you know. And so now I’ve gone from something that’s almost 800,000 people back to something that’s like six people. So a little bit getting my sea legs there, but it’s been really great. I sort of jumped in, started working with some of our clients already, and we’re pushing the envelope on some of this stuff that we’re doing in semantics, especially around things like virtualization, federation, and really serving up a semantic backbone for things like AI.
Larry:
Very cool. It’s like what everybody’s kind of doing now, stuff in that area. There’s so much demand for this. It’s really gratifying to see it. But I’m really curious about for you as a practicing ontologist and consultant, you were doing, as I recall, you were doing more kind of standard, not standard, but industry stuff and more like… Whereas knowing Tom and you and K3, are you focused on pharma and the life sciences in this new practice?
Eric:
We are currently, and a lot of that is because we’re small and we want to be focused. So we both have a really strong background. Tom, probably stronger than mine even in this, as well as our CTO, Ivan, Ivan Stankov is another guy who comes from the AstraZeneca background. So those guys have been really, really deep into pharma. I had been a consultant in pharmaceuticals. I’ve been a consultant in medical device. I’ve also though been a consultant in everything like oil and gas, finance, consumer packaged goods.
Eric:
I spent a lot of my career in the defense industry. So I used to be one of these guys with a top secret SCI clearance doing a lot of three-letter agency work and stuff like that, building ontologies around threats and stuff for data fusion applications. But Tom and I had this talk and we decided, “Hey, look, we both have a pretty good Rolodex in the space. It’s a good place for us to get started. We have a lot of deep subject matter expertise and knowledge in the space, so it makes a lot of sense.”
Eric:
We’re also both really, really into fair data, making data both findable, accessible, interoperable, and reusable. And those standards are really prevalent in the life sciences space. I mean, if you look around life sciences, there’s something like 1,100 plus ontologies to grab in the life sciences space, whereas you go to finance and you have basically FIBO, and you go to something like supply chain and you’ve got one or two. Right? And you’ve always just got a couple of papers on things.
Eric:
But so I think it’s sort of by design, but also through a product of necessity. But that doesn’t mean we need to stay there. We do have desires to move eventually as we grow out of life sciences. But Knowledge3, as a technology, it’s not bound to the life sciences at all. I mean, we could apply this to virtually anything, but you do need good heavyweight semantics when you are doing this kind of stuff in life sciences. So that’s another reason we want to kind of get in and solve these problems because we think it’s a really good starting point. And I think a lot of the companies in life sciences could really use the technology that we’ve got and the way we’re approaching this.
Larry:
Yeah, cool. It sounds like a really good focus for you. And it also reminds me of the way I originally connected about doing the podcast was your talk at KGC on super domains. And as you were just talking about this variety of life sciences ontologies to deal with, I assume you’ll be doing a lot of that, but I’d love to just, for people who weren’t at KGC, can you just sort of give us a quick audio overview of the concept?
Eric:
Yeah, yeah, no problem. So super domains is an interesting concept. In fact, I co-authored a paper now with-
Larry:
Oh.
Eric:
… Dr. John Beverley at University of Buffalo, and it’s been accepted into the ONTOBRAS Conference, which is the Ontologies in Brazil Conference coming up in September. So we will have a publication out on this that will be a short-scale paper around six pages. But the concept goes kind of like this. A lot of people, they want to use ontologies to solve a lot of complex problems.
Eric:
And in order to do so, a lot of what needs to get done is you need to run reasoners against these graphs. So it’s one thing to build a knowledge graph. It’s another thing to have really nice queries to query the knowledge graph, to integrate your data and do these things. But really what people are looking for is can the knowledge graph provide you with some kind of added information? Can you generate new kinds of triples? Can you generate new graph patterns out of this thing?
Eric:
And when you do so, the problem always was, well, where do I put those new triples? Where do I put those new graph products that are the product of reasoning? And the thing is they become very, very contextualized. So let me give you an example. If I’m in a bank and I want to know who my highest net worth individuals are, and I generate that graph using some rules and some reasoning, and I find people whose accounts are at certain levels and who have certain kinds of job titles and who may have certain kinds of investments.
Eric:
So there’s a whole pattern of someone I might look for who’s a high net worth individual. I may look for something like what are some highly prevalent adverse events in a certain class of drugs? Right? These kinds of things. These are context graphs. They’re not going to hold all the time. High net worth individuals will change from week to week, quarter to quarter. Adverse events around drugs will often change from study to study, new ones pop up. There’s a lot of things that we have to deal with that these reasoners produce.
Eric:
And the question is, where do you put that graph information? So traditionally, a lot of people who’ve worked in this space will know that you oftentimes choose between doing reification or now the newer version, which people are proponents of called RDF-star. Both of those are just fine. We’re not saying don’t do that, but they are kind of wrought with some dangers. Right? That we’ve seen over the years. One thing you get with reification is you start to put comments or triples on existing triples.
Eric:
So now I have a triple, a subject predicate object relationship in the graph, and I want to say, “Well, this thing only holds for males.” So I put a triple on that says “This triple only holds for males.” And then I want to say, “Well, it actually only holds for males in a certain age range.” So I want to put that triple on as well. And so after I keep reifying my triples, number one, my graph gets really, really large because if I do this a lot, I’ve seen graphs go up a thousand to 10,000 X in size.
Eric:
And the problem is your triples are bound to specific individual triples at this point. So again, what do I do if I generate an entire new graph? What do I do if I generate an entire new pattern? Okay? I have to go and tear those triples apart and put them onto specific triples in the graph. What a superdomain says is, “No, you don’t need a new language. You don’t have to do anything special. It’s more of an organization principle around good engineering practices.”
Eric:
So instead of talking about domains and just sub-domains that are the domains underneath them, I thought, “Well, what if we call these super domains the domains that are these context-driven domains that exist above the domains? And we put them off to the side?” Right? So it’s you literally move the information in a super domain to a different location. You give those graphs a name. Right?
Eric:
So you actually use the naming of the graph, use the URI or IRI pattern of the graph, and you can use hashes or slashes or however you want to do it, but give it a unique address and capture the context in that name of that graph if you want, or at least put it in a space and write some definitions around it like, “This graph only holds for males in a certain age range who take a certain kind of drug or who have a certain kind of disease,” or “This graph only holds on the first Tuesday of every month,” or “This graph only holds for individuals for the next three months of the calendar year, and then all bets are off.”
Eric:
So it allows you to have these really transient graphs, put them somewhere special, give them a name, and then given RDF 1.2 standards. I can take that entire super domain graph, refer to it back as a single node in my ontology, and now I have a really clean way to handle all these context graphs I might be generating over and over and over. And it just gives me a wing of the ontology to put them in so that they’re not intermixed with my base graphs, which I want to be true all the time.
Larry:
Interesting. I hadn’t thought about the implications of RDF 1.2 because that gives you a tidier place to put them, I guess. Because I love the way you… In fact, if it’s okay with you, I’ll put the slide from your talk.
Eric:
Sure.
Larry:
Because they were literally, first you talked about super domains generically as just the kind of layering, and then you just slapped the super domains on the side. One thing I want to mention that I thought seemed really relevant, I think you mentioned this in your talk or maybe in a prior conversation, is that those can be as fleeting as just something that’s there for literally seconds in some transactional thing or something like that. Or it could be some compliance-driven need to keep clinical trials data for 10 years or something like that. So-
Eric:
Correct.
Larry:
… so the concept holds that they’re temporary things, but temporary can be a long time.
Eric:
Yeah, yeah, and that’s right. I mean, if you think about it, you may want a super domain around ATM transactions that are happening only for seconds to minutes. You may want a superdomain to hold information by quarter. If you’re looking again at your bank and you’re looking at high net worth individuals. You may want a superdomain to hold for an entire clinical trial phase. Right? Or even across multiple phases, which means it could have to hold for years.
Eric:
But the idea here is that another reason that this is gaining some traction, and I have a colleague, Umesh Bhatt, who’s actually been putting this through some paces, if you will, using it on medical literature review, because… Here’s another good example of where these things become useful. If we imagine we’re going to put a knowledge graph underneath our AI. Right? To give it some kind of a foundation. Well, what happens with AI when you start to give it these medical review or medical research papers? Let’s say I have a paper and I say, “Well, look, I know that the gene IL-5 is responsible for asthma.” Right? It’s a biomarker.
Eric:
Not responsible really, but it’s a biomarker for asthma. Right? So it has definitely something to do biologically with asthma because it’s a biomarker for basically inflammation in the respiratory system. Okay? So if I write a research paper though, and let’s say I’m a researcher and I want to ask, “Well, could IL-5 also maybe be responsible for inflammation in the renal system, like in your kidneys or something like this, or be responsible for inflammation in your joints or something like this?”
Eric:
And so I write a paper and I put some hypotheses in there and I say, “Look, there’s some interesting evidence that maybe IL-5 is a good biomarker for some other kind of disease.” When the LLM goes over that paper, it will oftentimes get very confused. We’ve seen these things go into these papers and come back and say IL-5 is a biomarker for asthma, and IL-5 is a biomarker for renal disease, and IL-5 is a biomarker for inflammation in your joints. It could be for gout or something like that.
Eric:
So it’ll make these mistakes because it has a really hard time understanding facts from conjectures or facts from hypotheses. And what super domains are really about is it’s a hypothetical graph. It’s a conjecture graph. It’s a graph that could be ultimately true for a time period. Could also be a graph that isn’t really true. It could be a graph that is only maybe true. Right? So the idea really, Larry, came from back when I was working in the intelligence space. It was when I was doing government military intelligence because we dealt with a lot of what we called stochastic data, meaning data that is not factual.
Eric:
It’s noisy, it’s uncertain. And so when you start to work on threat conditions, who’s a good guy, who’s a bad guy, is this person part of a terrorist network or not, you oftentimes were dealing with very imprecise data and you had to generate hypotheses. And so you would get these hypotheses that you had to put somewhere because they’re useful. I may want to use this hypothesis to build another hypothesis. I may want to link some of these together with some of my known facts and see what comes out of this. Right?
Eric:
But again, I don’t want to put the hypothesis in my base graph or else my ontology gets very confusing very fast. Right? Because now my IL-5 biomarker isn’t factually connected to asthma. It’s potentially connected to other things. I just don’t want to commit to those other things right now. And so by putting it over there and giving it a different kind of a name, giving it an address and saying, “Look, maybe this is only true on Taco Tuesday, or maybe this is only true for three months, or maybe this isn’t even true.
Eric:
Maybe it’s just only partially true, or it’s just a hypothesis I’m kicking around right now.” Those are the reasons for super domains. And so a lot of people nowadays are talking about context graphs, and this is just another way to think about using context graphs, but keeping it out of the models where you care about the real agreed upon facts in there. Right?
Larry:
Yeah. I was going to ask about that because you mentioned that earlier that these are essentially context graphs. And I don’t think I had not heard that term used until the Foundation Capital article last December. And then everybody, that’s all anybody talked about. Did you call them that before then or?
Eric:
Yeah. So I’ve been thinking, so again, not being a computer scientist and being an ontologist that comes out of philosophy, I think I’ve used a lot of weird words for a lot of years. I’ve got clients that would admit to this that I was the guy sometimes walking in and talking about the doctrine of hylomorphism, which is how matter and form are inherently connected in Aristotle’s metaphysics. Right? So the fact that any time that you see something, you see it’s kind or it’s species. Right? I’m looking at my computer screen right now, my laptop screen.
Eric:
So I’m not only seeing my specific instance of a laptop screen, I’m also seeing its kind. I’m seeing a type of thing as well. And so for me, context, things like using German words like Sachverhalt for states of affairs or using things like contexts or using things like environment or environmental type of graphs or talking about things in terms of different types of logical structures and such. Yeah, that’s all been kind of I sort of think worn, sort of woven into my vocabulary over the years probably because I’m always approaching this from the perspective of metaphysics and not so much the perspective of just computer science.
Larry:
Yeah, I love that. In fact, I think it was right before your talk, there was a panel I think or something.
Eric:
Yeah.
Larry:
And a little philosophy debate broke out on stage. It was great. It was you or it was Jamie McCusker and who was it? And you, maybe Casey Hart or somebody or?
Eric:
Yeah, there was someone from BMS. There was a senior guy from BMS, I forget his name now, and me, Jamie. Yeah. And I think Deb was on there too. I think Deb McGuinness might’ve been on there.
Larry:
Right. I think she was chairing the room that day.
Eric:
Yeah.
Larry:
Yeah.
Eric:
And so yeah, that one got into some philosophy stuff because we started to talk about this notion of graphs and ontologies, and somebody said something about context graphs. And the thing that always kind of gets to me a little bit is ontologists who are computer science trained, they don’t always make a clear distinction between ontology, which is in Greek, the study of existence, and epistemology, which is the study of knowledge. So a lot of times people talk about the Tom Gruber definition of an ontology as a specification of a conceptualization. You might remember that one.
Larry:
Yeah. That’s the one everybody says first. Yeah. Yeah.
Eric:
Yeah, but that’s epistemology. Right?
Larry:
Oh.
Eric:
A specification of a conceptualization means I’m specifying a thought pattern I have. I’m specifying something I’m thinking about. In my world of growing up reading Husserl and doing phenomenology and these kinds of things, that means I’m really talking about my mental act. Right? Things about my mental act, not necessarily things about the physical object in the world that my mental act is aimed at. It might even be a hallucinatory act, like the thing I’m thinking about.
Eric:
Maybe I’m thinking about the tallest unicorn or the most golden mountain, right? Things that don’t really exist, but that I could imagine. And so for me, I like to always be careful to separate the ontic or the things that are real from the epistemic or the things that are mental acts or thought processes. Now, the two things definitely intertwine. A lot of our models revolve around capturing knowledge from people, sure.
Eric:
But I’m a faithful realist ontologist. I think our conceptualizations are tied to our biology, they’re tied to our brains, our physicalism in our brains. They’re tied to the world and the objects around us and the things we perceive. So being a phenomenologist, I’m always careful about separating my acts from objects. And I see that get confused sometimes. So that’s where I have to take responsibility. That’s probably where the philosophical skirmish broke out from.
Larry:
Yeah, I love that. I just love that that’s a thing in this field, that there’s enough people around that that kind of thing can break out. But also, I think it points to in terms of practice, I don’t know, there’s a lot of trained philosophers, like logicians and ontologists.
Larry:
I’m just thinking of everything from you to Casey Hart and even Pat Hayes who worked on some of the original RDF and OWL stuff. He’s a trained logician, things like that. Has there been sort of a meeting of the minds? Was that just sort of a philosophical side excursion or are there fundamental practice things that still need to be worked out in terms of properly sorting out epistemology and ontology?
Eric:
I think there’s still a lot to sort out there realistically. I think there’s some careful things to do there. I also think though that some of it is about the perspectives and then some of it is about getting real things done in the real world.
Larry:
Yep. Yeah. No, that’s the other thing I love about this community is the level of pragmatism and user and business focus that people bring to their work. Hey, but kind of to that last point about both your story so far, leaving consulting to start this new company or join this new company with Tom, and then your approach and this kind of methodological thing you develop with super domains, you’ve been in this business for a while. And one of the things I like to keep on top of in this podcast is where are we at as a field of ontology practice?
Larry:
No, but what I was saying is that your story so far, this path, you’re kind of leaving consulting to join this smaller company, it sounds like that’s a step back into practice and stuff. And then the stuff you were just saying about the practice evolution you’ve made, this articulation of super domains and how they work, it just makes me think that there’s so much going on right now in the field and that I wonder how much your moves are a microcosm of what’s going on or just… Because I think of for the example, the way LLMs and the GenAI stuff has affected what we do. What’s your take on where are we at as a field right now?
Eric:
Yeah, that’s a good question. I mean, I really think that this is a great time for semantics. And the reason is because I really think that right now people finally don’t have the choice anymore to kick the can down the road on their metadata problems. So having been in this space a long time, I’ve seen it be influenced a ton by a lot of other technologies. So if we think back. Right? I mean, ontology started getting big when XML first hit the scene. And so there was a lot of push like, “Well, XML can handle a lot of this stuff.” Right?
Eric:
And then you had the notion of, well, ontologies are great, but then big data happened. Okay? And then it was like, oh, well, do we need graphs if we can have these massive tables and we’ve got Hadoop now and we can do the… But the problem with big data was you still needed an index. Right? You still had to put the data together somehow, and that was the hard part. So the thing was a lot of people over the years in a lot of industries have constantly just kicked the can down the road for these types of metadata problems that they were facing.
Eric:
Well, now with AI, I think the investment in AI is so big, it’s happening so much that people have to get this metadata part under control. And the reason I’m saying that is the way that AI works is it’s picking paths through a vector database based completely on statistics and probabilities. And this means that number one, these things don’t really know how to handle logical arguments or logical distinctions very well. Apple showed this recently in their illusion of thinking paper.
Eric:
There’s also an issue of trying to get an LLM to repeat a task over and over and over is pretty close to impossible because it seems like they just kind of get bored. They want to do it a different way. They want to come up with a different path through the data. Right? Because every time it’s taking a prompt and it’s thinking about it, and it’s going to come up with a slightly different variation to solve it or a slightly different way to answer the question.
Eric:
I’ve even played just with imagery stuff where I give it a picture and say, “Okay, I want this exact picture, but just change this one tiny little thing.” And it will redraw the picture. It’s close, but it’s not exact. Right? It just kind of redraws the whole image. So what does that mean? That means that AI really lacks a good deterministic backbone. It doesn’t do well when it comes to mapping into structured types of data. It doesn’t do well when you say, “Go fetch me the same pattern over and over and over.” Right?
Eric:
“Fetch me the same data from the same pattern and run the same process again and again and again.” Right? Or when you have this context stuff, but you want to have some facts that don’t move and then just add some context on top of it. Right? Because of those things, that determinism in knowledge graphs, the way that they map, the way that they can model metadata, the way that they can organize data, the way that they can use standards, the way that you can deductively fix your logic or even use some inductive logic is fundamentally different than what the AI engines are doing.
Eric:
So everyone has decided we’re going to make a big investment into AI platforms. Great. What it does, it does really well. It reads really well. It summarizes really well, and it can write really well. I don’t think it thinks very well. And so if you can put this determinism behind this, you can actually start to make AI, number one, way more trustworthy. Number two, you can avoid these kinds of hallucinations that we see all the time. One of the problems I have with hallucinations is that in systems we used to build a long time ago in the intelligence community, everything had to come with a confidence score.
Eric:
You didn’t just give an answer like, “Yes, go hit that target” or “Do this” or “Do that.” It was, “I am 0.78 sure that you should hit that target,” or “I’m 0.96 sure that that’s the same thing as this, or that this person is a member of this organization” or something or whatever. And so you needed those confidence scores because kind of that told you, the system was saying basically, I’m pretty sure about this or I’m not very sure. AI to me, it’s usually pretty confident in itself. And if you challenge it, a lot of times it just kind of gives you a new answer because it’s also trying to please you.
Eric:
But I mean, ultimately, we’re trying to give this determinism behind AI so that you can cut down on these things. Right? You can make it more trustworthy, no black box answers, limit the hallucinations, all these kind of things. So in order to do that, I think there’s been significant advantages in showing that using graphs behind RAG, using a graph RAG approach for AI is a good way to do this.
Eric:
I think that the best quality graphs you can build, right? So having graphs that have some of this deeper semantics that understand what we were talking about before, like your epistemology versus your ontology, having a realist view of the world, being able to manage your context graphs from your fact-based graphs in things like super domains.
Eric:
If you put all of this stuff together, to my mind, I mean, this is what’s needed if we’re going to make AI work at scale and if industries are going to use this stuff in a way that they can actually trust it and make real bets with it. And so if the investment’s being made and we’ve decided graphs are a good thing to do here and we need this determinism, then I think this is a great time to be an ontologist and it’s a great time to have this tool in your toolbox.
Larry:
Yeah. No, it’s definitely feeling like that. And yeah, again, it’s so gratifying to see that. But hey, Eric, I can’t believe it, Eric. We’re coming up close to time already. And that was a great. I loved what you just said. That’s such a great summary of kind of where we’re at. But before we wrap up, is there anything last, anything you want to revisit from the conversation or just want to make sure we share before we close?
Eric:
Well, I think the last thing, not getting too corporate or pitchy or anything here, but I mean, one of the reasons when we talked about me coming to something like Knowledge3 was not only to work with some friends and colleagues that I have ultimate respect for and that we see the world in a lot of the same ways. I think a missing piece in a lot of this, Larry, is there are no real products out there that are doing the work that people need done. That to me is what was important.
Eric:
I saw that at Accenture a lot. Accenture has some very, very good methodologies, very good ideas to bring their clients forward, but a lot of clients were always asking for these accelerators. Right? They really wanted something that they could put in their environment and turn the crank, and it started to make some of this easier. It started to make it a bit automagic. Aright? And so what you have in this semantic space, unfortunately, over the last 30 years that this technology’s been around is you still have to basically buy a lot of piece parts.
Eric:
You have to buy an ontology editor tool to build your ontology and write it all down in. You have to buy some type of a graph database like a triple store or a labeled property graph store or something to put your graph in. You have to build some kind of an API framework. You have to maybe have a natural language processing engine of some type, or now you need an AI agent or an agentic framework or some bots that are going to do things. So you wind up with all these pieces. And the problem, of course, is people just want to buy a car or a motorcycle.
Eric:
They don’t want to have boxes of parts show up in their garage and have to build a thing. Right? You want to get in, you want to turn the key, you want to drive somewhere. Some people, I’m a motorcycle guy. I would love if boxes of parts showed up and I could spend a year building a motorcycle. That’d be a lot of fun. But that’s not what I need if I have to get something… If I need something to get to work tomorrow, that’s not the answer I want.
Eric:
So I think a lot of companies now are in this position where they’re saying, “Okay, I get the theory. I understand we need this determinism. I understand we need these piece parts. How do I get there? And how do I get there at the speed of AI?” That’s a slogan I’m coming up with at Knowledge3 is I’m talking to people about what you need is you need semantics at the speed of AI. Right? And what I mean by that is you can’t wait for a year or two to build the knowledge graph engine that’s going to put the guardrails around your AI because in a year or two, you’ll have lost the race to the AI.
Eric:
Right? There’ll be a new AI, there’ll be something else. I mean, your competitors will have lapped you. So instead of having these pieces and then thinking about, “Well, I don’t know how to build an ontology in an editor,” or “I don’t know how to store it in a knowledge graph,” or “I don’t know how to write SPARQL queries,” or “I don’t know how to do any of this stuff.” What Knowledge3 is all about is we are building all that connective tissue. We are going to work with the other vendors, we’re going to work with the editor companies, with the triple store companies.
Eric:
We’re going to work with the API companies, with the cloud hyperscalers, with the different AI engines. We see ourselves as basically taking those parts and giving you the whole organism. You get all the connective tissue. Right? All the organs will be able to work together. The blood’s going to flow. Right? The receptors will all be there. Right? So it’s this kind of thing like, that’s what I’m seeing people really wanting is they kind of want an easy button to push so that they can get going. And that’s really what we’re trying to do.
Eric:
We’re trying to tell people, “Listen, if you go with our stuff, I don’t have any magic answers. There’s no free lunch. You still have to build your models, you still have to get your metadata right, you still have to get your competency questions right.” But if you had an engine that could speed this up where you’ve got a minimum viable product, you’ve got something working in weeks to a couple of months versus several months or quarters to a year or two, that path of acceleration I think is really exciting right now in this space.
Larry:
That’s very cool. And it sounds very needed right now because the on-ramp to the RDF stack is notoriously a little challenging, but… Well, that’s very cool. It’d be fun to watch how you all develop. But hey, one very last thing, Eric.
Eric:
Yep.
Larry:
If folks want to follow you, what’s the best place to connect online or to follow you?
Eric:
You can hit me up on LinkedIn. I’m usually pretty good about accepting connections, responding to people there. I don’t really do a lot of social media anymore. I used to be on Twitter and that, I gave that up because of how that’s gone. I do have a Bluesky account, not really on it very much. But normally I like LinkedIn, find me at conferences. People can email me if they want to. If they really have some questions or if they want to figure out how to get in touch with us. So yeah, just basically those kinds of things.
Eric:
Keep an eye out for what we’re doing. I plan to be publishing and putting out more content now, so I’m hoping to get some interaction and engagement. And we are looking at setting up a Knowledge3 Academy where we’re going to put some of our content out as lectures. We’re going to have more of an interactive way for people to come in and kind of learn things and understand some of these fundamental concepts that we’re talking about. Because we’re really trying to educate the market and provide them with tools and techniques that can help them go faster.
Eric:
And we don’t need to be the ones to own all this. We see ourselves staying pretty small and boutique. I would really just like to give people that missing cog, if you will, in their machine that they can put in and it’ll help turn all the cranks that they need. They can come to us for kind of the hard questions and such. We can get them going.
Eric:
And then let’s train them up and let people take this stuff themselves, own it, run with it. I mean, the world’s full of a lot of smart people. They just don’t always have the right kind of training in school to do this. So we’re kind of looking to upset the market a little bit in that way. I think we can maybe just get people along in a way that they’ve not been able to sort of advance in the last few years.
Larry:
Well, that approach resonates. I don’t know if I told you, my whole intent in this podcast is to responsibly democratize all of this practice, and I applaud anybody else who’s trying that, so good on you.
Eric:
Thanks.
Larry:
Hey, well, this has been great, Eric. Thanks so much.
Eric:
Awesome. Thanks, Larry. Thanks for the invite. I’m looking forward to seeing this and big fan of what you guys do in exposing this community. So it’s great. It’s been awesome. Thanks a lot.
Larry:
My pleasure.
