Podcast: Play in new window | Download
Subscribe: Apple Podcasts | Spotify | Amazon Music | Android | Youtube Music | RSS
Every profession has its connectors, sharers, and community organizers. In the knowledge graph world, George Anadiotis fills all of these roles.
Through his industry analysis and reporting, his conference organizing, and his writing and podcasting, George connects ideas and people across the semantic-tech landscape.
We talked about:
- his work at Linked Data Orchestration and as a consultant and analyst in the knowledge graph and linked-data world
- his diverse background in computing and his studies at the intersection of knowledge management, the semantic web, and distributed systems
- his extensive writing experience and consulting background
- his definition of a knowledge graph
- the differences between RDF-based knowledge graphs and labeled property graphs (LPG)
- the focus in the RDF community on standards and interoperability versus the focus in the LPG community on implementation
- the variety of query languages in the LPG world and recent efforts like GQL to create a standard way of querying LPGs, as well as efforts to query across both RDF and LPG graphs
- the origins of his annual Year of the Graph report
- some of the reasons that knowledge graphs are positioned in the bullseye of Gartner’s Impact Radar this year
- where knowledge graphs fit in the AI landscape
- the role of knowledge graphs in RAG architectures
- the conference he organizes, Connected Data London, coming up December 11-13
George’s bio
George Anadiotis has got tech, data, AI and media, and he’s not afraid to use them.
He helps organizations map and understand complex domains to make better decisions; design, implement and monitor models, processes and systems to achieve goals; and craft communication strategies and outreach initiatives to grow awareness and market share.
He enjoys researching, developing, applying, writing and talking about cutting edge concepts and technology, and their implications on society and business.
Connect with George online
George’s publications, podcasts, and conference
- Connected Data London (conference roundtable recording)
- The Year of the Graph
- Orchestrate All the Things
Video
Here’s the video version of our conversation:
Podcast intro transcript
This is the Knowledge Graph Insights podcast, episode number 3. In any domain, there are people who seem to do it all – practice and consultation, industry analysis and reporting, and community building and event organizing. In the world of knowledge graphs and the semantic web, George Anadiotis has filled all of these roles. Whether he’s publishing his Year of the Graph newsletter, organizing the annual Connected Data conference, or producing the latest Orchestrate All the Things podcast, George is always connecting the dots.
Interview transcript
Larry:
Hi, everyone. Welcome to episode number three of the Knowledge Graph Insights Contest. Sorry, I’m going to redo that again. I have too many podcasts. I need a new intro for this one. Okay. Hi, everyone. Welcome to episode number three of the Knowledge Graph Insights podcast. I am really delighted today to welcome to the show George Anadiotis. George is really well known in the knowledge graph world and the graph world in general and the tech world in general, as an analyst, a consultant, a really well-developed engineer. He runs a big conference around knowledge graph and graph technology, and he is the principal at his organization called Linked Data Orchestration. So welcome, George, tell the folks a little bit more about what you’re up to these days.
George:
Great. Thanks for the intro, Larry, and good to be here. Actually one of the opening, I guess, guests for this new podcast series of yours. Well, the truth is I have a long and kind of convoluted story, but I’ve kind of honed my skills of telling it in as simple way as possible. So basically in terms of background, I have a very hardcore computer science background. I was one of those kids that I saw my first computer when I was like 12, and immediately I kind of snapped and I realized, “Okay, so this is what I’m going to do in life.” So went to college, studied computer science, graduated, started working as a software engineer and architect and all of that stuff, consultant, all of that. And then at some point, about a decade in basically, I realized that it’s been fun, but I wanted to try something new.
George:
And that’s the point where graphs sort of entered my life because the thing… I was interested in research and my topic was somewhere around the intersection of knowledge management, semantic web and distributed systems, and there was a specific group that I wanted to join that was working precisely on the intersection of those things, the Knowledge Representation and Reasoning group based in Amsterdam, led by one of my mentors, Frank Van Harmerlen. So I was lucky enough to spend a few years there, did some really cool stuff in that group up until the point where I left, I repatriated. So I should also mention I’m from Greece originally, so I spent a few years in Amsterdam, then moved back to Greece, kept working on the intersection of those technologies actually. But for a few years I did that leading the R&D of a company that was developing projects and products around those up until 2012.
George:
And that’s the point where I started doing my solopreneur thing. So ever since I’ve been juggling a few things. So I work as an analyst, I collaborate with GigaOm, I work as a writer. I’ve contributed to a few publications such as VentureBeat and ZDNET. I have my own newsletter and blog called the Orchestrate all the Things podcast and newsletter. What else? Let’s see. As you mentioned, I organize an event, it’s called Connected Data. I think we can elaborate a little bit on that later because it’s actually very much relevant for the knowledge graph theme. I also curate a newsletter called The Year of the Graph and their graph database report also going by the same name to consulting with a number of companies from… Pretty much everything ranging from go-to-market strategy, marketing to technical implementation. So I juggle many balls, as I said.
Larry:
I’m exhausted just listening to that and I feel very grateful that you found the time to talk to me with all that you have going on. So thank you, George. Hey, one thing I like to start each episode with is I would love to get your definition of a knowledge graph just for folks… The idea is to hopefully come up with something like a canonical definition somewhere in the next one to five years. But anyhow, I’d love to get your take on what a knowledge graph is.
George:
Yeah, that’s a good one. And somehow it never gets old. I’m sure you’re probably familiar with the fact that I believe a couple of years when I last checked, I think there were over 100 definitions for what constitutes a knowledge graph. So they’ve probably grown to, I don’t know, maybe 200 by now. So 200, 201 who’s counting. I’m going to give you mine as well. And by the way, if you ask me next year, I’m probably going to tell you something slightly different, but here’s the current definition. So if we’re talking about the graph, then it basically means that we’re talking about the data model in which the key elements are nodes and edges.
George:
I’m going to add the directed adjective to edges because well, if you only have edges without direction, it may lead to some ambiguity, let’s say. So that’s the graph part, which it’s not very original, it’s kind of the textbook definition. What may be a bit more original is the knowledge part. So I think in order to be able to qualify a graph as being a knowledge graph, I think there are certain conditions that need to be met. So basically I think that both nodes and edges should enable users to define their properties and they should adhere to a schema. That’s as lightweight as I could possibly keep it without getting too technical.
Larry:
Interesting. And that notion of a schema and both having properties on the edges. I guess maybe want to diverge just a little bit and talk about the difference between an RDF-based, triple-based knowledge graph and a labeled property graph. Can you talk a little bit… That might be another thing to really get… Because I think a lot of people, when they hear graph technology, they’re thinking most… I think the most common databases and tools that are out there are often around labeled property graphs. So can you help us tease out between an RDF-based knowledge graph and a labeled property graph?
George:
Okay, well, we could spend at least four or five podcast episodes talking just about that. And by the way, there recently was another episode by a good friend actually, and also very knowledgeable person in the graph world, Amy Hodler. So she spent an entire episode with her guests dissecting this exact topic. So what’s an RDF-knowledge graph? What’s a labeled property graph? How are they different? When should I use what and so on. So I’m not going to even try and be as extensive as they were, but let me just say that for most people, if they’re not familiar with graphs at all, maybe just the general idea of the graph data model, let’s say, they don’t even know… They can’t actually imagine, I’m guessing, that in the graph world you do have this kind of schism.
George:
So there’s two ways of modeling graphs, because if you think about it, that’s not the case for relational data. As far as I can remember, that’s not the case for the document data model either. So in those words, things are pretty straightforward. Okay, so I want to build a relational database. I have tables, I have SQL. There’s one way to go, basically. Yes, you may have slight variations on the query languages, but for the most part it’s all pretty standard and pretty straightforward. That’s not the case in the graph world. So yes, you’re right. There are two, let’s say, major flavors of graph data models, RDF and labeled property graphs.
George:
So, like I said, I think other people have done a better job than I could possibly have in a couple of minutes explaining what the similarities and the difference are. Let me just say that historically, I believe the reason why we have two data models is because these data models have originated from two different communities. So the RDF data model basically comes from the semantic web world. So historically, this has been mostly academic, mostly defined top level from certain experts and mostly with the intent to facilitate knowledge management use cases and semantic web use cases. So when you get these set of assumptions, well by the outcome, this is what you get. Something like RDF.
George:
On the other hand, the labeled property graph data model comes more from the software engineering world and more from people who were interested in a different set of use cases. So they were more interested in closed-world assumption, working with analytics and OLTP use cases and building applications. So when you start with this set of assumptions, well you end up with something like the labeled property graph. I’m not going to say, “Oh, this is better,” or “That one is better” because it doesn’t lead anywhere. Let’s just say that they’re different. You should be aware of their differences and depending on what you want to do, well, you choose one, unfortunately, I may add. But fortunately, there are also multiple efforts underway to bridge those two worlds, and there are already some good ways that enable you to sort of cross over.
Larry:
Yeah, interesting. That just what you just said I think is very… Especially since you’ve sorted out the use cases that drive adoption of those two choices, I think that’s super helpful right there. And that’s probably how most people come into it. You just mentioned the efforts that are underway. I’m thinking like GQL was just now really… I don’t know if that has to do with it, but what are the things that you see going on that might lead to more connection between the two models?
George:
You’re right. Actually, GQL may well be one of those things. And I should actually pause a little bit here to explain what GQL is because I’m guessing that well, many people won’t have a clue. I mean, just another acronym. So GQL stands for Graph Query language. And well, since we started talking about the different data models, so RDF and labeled property graphs, one of the differences has historically been that, well, RDF having this kind of background and having been around for over two decades by now is standardized. And historically this has been one of its greatest strengths because what it means is that, well, you have a wide variety of choice in terms of tools, platforms that you can use to work with RDF knowledge graphs, and the fact that it’s a standard means that you can easily switch from one to another and things will for the most part keep working very, very smoothly. You can import and export data. The query language is also standard. So that’s been a great facilitator in terms of working with the RDF model.
George:
Now in the LPG world, things have not been as fortunate up until now. And again, I do believe that has a lot to do with, well, it’s origin. So like we said previously, the LPG model is something that has evolved through implementation basically. So people who were less worried about standards and interoperability and more focused on delivering something that works for their specific use case as fast as possible and tweaking it to their needs. That’s good. I mean, in some ways, but one of the downsides is that by taking this approach, you end up with multiple ways to model knowledge graphs using the LPG model, and you also end up with multiple query languages. And that’s exactly what things have been like well, since the onset of LPG basically up until a couple of months ago recently, I guess.
George:
Precisely because the people who were involved in building those solutions as well as the people who were using these solutions realized that this is a problem that’s hampering adoptions, eventually they decided to sort of put their differences aside and just all gather around some standardization effort and try to pick and choose the best parts from each query language and come up with a new proposal for a universal query language for the LPG world. And that language has been evolving since 2019. I should also mention, by the way, that I was one of the few people who were lucky enough to actually be there at the beginning. Initially this initiative came about through a W3C Workshop that was held in Berlin in 2019, and I was invited to that workshop along with some other prominent people from the knowledge graph world. And the idea there was basically to get everyone around the table to form working groups and to establish some patterns and some principles that would enable people to start working towards this goal, common query language for property knowledge graphs.
George:
And five years later, and I believe it was end of February or early May, something like that, this effort finally came to fruition and the GQL standard was published by the ISO. So it’s a standard now. It’s official. What’s left? Obviously, in a way though that’s only the beginning because vendors have to implement the standard. There has to be education, adoption and all of those things, but it’s a great starting point. And by the way, this also means that… Well, it’s a good thing in and of itself because more standardization means better adoption basically as far as the LPG world is concerned. But as part of that effort, there have been some spin-offs, let’s say, from people who saw this as an opportunity to go even further and try to bridge the RDF world with the LPG world. So there’s a couple of these efforts going around at this point, and I’m fairly optimistic that this may provide a starting point, a springboard for even further convergence.
Larry:
Nice. And it all starts with a standard too, the fact that… And it sounds like those folks will be building on the standard, so there would still be that foundation that you all started five years ago. Hey, George, I want to shift gears a little bit because the reason I first contacted you about getting you on the show is your Year of the Graph report that you do. And I would love to hear… I was reviewing it the last few days and there’s so much in there. What do you think… I came up with at least 10 or 12 top level take-homes. Can you help me coalesce that into top, I don’t, three or four or five things that you think are really relevant to how graph tech and graph business practices will evolve in the next year or two?
George:
I’ll try to, it’s not an easy ask, but I’ll try anyhow. But let me just maybe share a little bit of background on what this is and how it got started and why it’s still around basically. So I started doing that in 2017, and at the time I was a contributor for ZDNET, which some of the people who may be listening to the conversation may be familiar with. And I was covering the general, let’s say, data/data science/analytics/machine learning AI world. And I was relatively fresh at that time, and somehow I was approached by a vendor active in the graph database world. And I thought, “Well, okay, why not? I’ll try writing an article about that.” And to my relative surprise, it was extremely well received. It was super popular basically. So I went, “Okay, so…”
George:
Because obviously, I myself have been familiar with the technology and have been using it in research and projects and all that stuff for a long, long time. But up until that point I considered it… To be honest with you, I considered it rather niche, but the kind of response I got made me start to think that, “Well, maybe the time has come for these technologies to get mentioned. And since I do have a background in this technology, there seems to be demand for it in terms of educational resources and just keeping up with what’s new.” It seemed like a gap to fill. And I was very quick and eager to fill that gap. So that’s when I started compiling the newsletter as a way, first of all, to keep track for my own benefit, let’s say, of what’s going on in the graph world. And then eventually it turned into something that… Well, it got a life of its own, let’s say. It has its own following. And so, I enjoy to keep doing that.
George:
So to come to your question, so what’s in the latest newsletter? I have to say, compiling the latest newsletter was the hardest. I keep saying that every time, but it’s true simply because of the fact that there’s more and more things going on, more and more things to take note of and to cover in that world. So when I started, I remember I was sort of struggling to find noteworthy items. Now I’m struggling to keep up, to be honest.
George:
So let’s see. I would probably start by the first item as I categorized it in the newsletter itself. So the fact that Gartner considers knowledge graphs to be a key enabling technology. So recently Gartner put out its latest report on emerging technologies, and it has an interesting sort of way to navigate that report. It classifies these technologies into different sectors and it has this sort of well, radar maybe or… I don’t know how you would call that visualization. Anyway, the point is that it ranks technologies on the basis of how central, how key Gartner believes these technologies are.
George:
And if you look at that visualization, there’s a couple of technologies that are right in the center, in the bullseye, and knowledge graphs are one of them, and there’s a little bit of an elaboration around that. So the reason why Gartner thinks that knowledge graphs is such a key technology is because, well, it is an enabler for pretty much everything else. If you have knowledge graphs in your organization, it means that, well, you have a structured approach, knowledge management, you have knowledge in a way that can be shared and reused across departments, applications, initiatives. So it really is an enabler. So that would be my number one item, not the fact that this is the case. I mean, this is something that anyone who’s been into knowledge graphs for even a short amount of time has quickly come to realize. But the fact that more and more people are realizing it’s getting very mainstream and it’s getting recognized, then that has its own interesting set of implications.
Larry:
And you mentioned the other thing at the center of that bullseye… And I know that impact radar thing well, because it’s now my desktop background on my laptop. I look at it every day. But the other thing at the center is AI. And one of the other things you talk about in the report is the relationship between graph technologies and AI. I know there’s a lot in there. What do you think are some of the key take-homes, especially for people just starting to get their head around graph technologies? And I think a lot of people right now are facing their boss coming to them saying, “Okay, we got to do AI,” and then they get into the implementation stuff. And I know there’s a number of places where graph technology can help. Can you maybe talk from that angle about how graphs and AI and the LLM stuff interact?
George:
Yeah. Yeah. Actually, let me start by saying something that I don’t think many people realize. I mean, knowledge graphs are also AI, by the way. It’s just that by convention, let’s say, these days when people say AI, they implicitly mean something like a machine learning approach or a large language model or something of this sort. So knowledge graphs are also AI. It’s just a different kind of AI, one that’s deterministic and one on which you can count on to deliver outcomes that are explainable and traceable and based on facts and all of that stuff. And that’s precisely the thing that makes them interesting in combination with LLMs and all of that stuff, because they’re complimentary really.
George:
So where LLMs shine is the fact that they’re easy to use and they’re quite creative, let’s say, in the sense that, well, you can specify pretty much anything. You can ask pretty much anything and you’ll get a reply to your question. The problem is that what you’ll get is not always factual. I mean, it’s the famous hallucination, so you are going to get something as a reply to your question, but you’re never sure whether that something is actually real or made up. And this is the part where knowledge graphs can help. Knowledge graphs on the other hand, come with their own set of challenges, probably the biggest of which is that they’re pretty hard to build and the expertise and the time and resources required to build solid knowledge graphs are quite considerable.
George:
However, if you have built that, it can complement something like a large language model very well. And this is the, so-called, Graph RAG Approach, RAG standing for Retrieval Augmented Generation. The main idea of which is basically, okay, so you can use your Gen AI, your large language model as a sort of user interface to express what it is that you want to get out of your system. Its language handling abilities come really handy for that. However, for retrieving the answer to your question, it’s best if you hook up your LLM to a knowledge base that’s actually factual and you can actually trust. And this is pretty much the definition of a knowledge graph. And this is how these technologies come together.
Larry:
Yeah. And that’s kind of all you see. At least all I see on LinkedIn these days is Graph RAG and RAG architectures of various kinds. That’s a whole other episode maybe that we could go into.
George:
Yes. A number of episodes.
Larry:
Yeah. One of the things that I know from the knowledge graph world is that one of the other differences… Well, the fact that there’s an ontology associated with a graph, I guess you can associate an ontology with an LLM as well, but the way knowledge is represented in a graph is… It kind of gets at what you were just saying about true factual things that human beings actually know. What are the key things if you’re building a RAG architecture, how do you… Because I know they almost always end as you said, that the UI is almost always the LLM because of their conversational nature and the way they work. How ubiquitous do you think the interaction between the two will be between the knowledge graphs and the LLMs in RAG architectures going forward? Because it seems like everything I read, there’s some new connection between the two.
George:
Yeah, you’re right. And well, I have to admit right from the start that I’m still learning myself, so I don’t have a definite answer to give you. What I can tell you is that people are coming up with different architectures and different names for those architectures as we speak pretty much. I mean, I just read an article, which is also included in the latest Year of the Graph issue, by the way, about classifying different RAG architectures a few days back. And I just read another one very much to the point by the way today as well. So there’s different parts in the RAG architecture, let’s say that graphs can be used, they can be used as a sort of metadata store, they can be used as a knowledge base, they can be used to… Well, they can even be used to store embeddings. So embeddings are basically… Embedding is a way of representing, let’s say, different facts about the world in whatever format. They can be text or the multimedia, whatever, in a way that machine learning algorithms can understand and somehow compare.
George:
So there’s all sorts of ways that graphs can be used in this architecture and well, like you said, it’s a topic in and of its own. And you may as well have another episode on that. And that’s also a good opportunity to plug something here because, well, I’m going to be moderating a panel about specifically Graph RAG in a few days from now. So on July 19, it’s going to be part of a small SWARM community event organized by Joaquin Melara, you, and perhaps some of the audience may know. So if you want to know more about Graph RAG, join us there. And I’m looking forward to that because I’m also eager to learn more.
Larry:
Yeah. No, Joaquin’s a good friend, and I’m really interested in the stuff he’s doing with SWARM. Unfortunately, this episode will air after that, but I will link to it and link to his stuff. In fact, there’ll be a lot of links in this episode, like Amy’s podcast you mentioned and of your report and all that.
Larry:
But hey, George, I can’t believe it, we’re already coming up close to time. I like to keep these around a half hour. But before we wrap up, is there anything left, anything you want to revisit from the conversation or just make sure that we share before we wrap up?
George:
Well, the one thing I would like to mention, which I don’t believe we’ve done so far, is that, well, if you’re interested in all of that stuff or even maybe some of that stuff, there’s an open invitation I have to people listening to the conversation, you should really come join us in London in December when the upcoming Connected Data London conference is taking place on December 11, 12, and 13. It’s the event I’m organizing with my partner in crime, as I call him, James Phare. It’s an event about all things, knowledge graph, graph AI, analytics, databases, semantic … all of the things that we’ve been talking about, and then some basically, and it’s going to be a blast.
George:
So three days full of master classes and presentations and keynotes and panels and workshops, an unconference. And it’s just too big to cover here. Let me just quickly point that if you want to learn more, we just had an online roundtable last week and it was recorded, and we just released the recording on YouTube today, so you can look it up for more. And obviously we have a website, we have social media, the information is all out there. We’re looking forward to people submitting their work and also joining to attend the event.
Larry:
Perfect. Yeah, I watched that roundtable and I’m super excited for the event. Really looking forward to that. And we’ll include links to all that as well. And I hope some of our listeners will submit papers or talks or… It sounds like in the scope you described, I’ll let people watch the roundtable, but it’s a really ambitious, rich program I’m going to observe. So really looking forward to that. Hey, one very last thing, George, if folks want to follow you online or connect and follow your feed, what’s the best way to follow you?
George:
Well, I’m on most social media, so they can look me up either by my name or by Orchestrate all the Things, which is the brand I use for my podcast and newsletter. So I’m on Twitter, LinkedIn, TikTok, even I started experimenting with that format a little bit, Instagram, you name it.
Larry:
Cool. Okay. I’ll put that in the show notes as well, and we’ll link to all that. Well, thank you so much, George. I really enjoyed the conversation.
George:
Great. Thanks for having me. And yeah, I mean, it went by really fast.
Larry:
It always does, but it’s an ongoing conversation. We’ll have you back soon…
