Wednesday, 14 June 2017

Podcast Interview with Sébastien Heymann, Linkurious

As I am coming up on my 5th anniversary working for Neo4j, I am increasingly happy, proud and thankful for the journey that we had - and the many great people that I have met along the way. One of these people is FINALLY appearing on this podcast, and has a history with this blog every since the VERY first article that I wrote in january 2013: in this article, I showed folks how to load the Belgian Beer Graph into Neo4j using a tool that was actually not intended for this use: Gephi. Many beer (related article)-s later, I am now finally talking to Sébastien Heymann, founder and CEO of, and one of the main people behind Gephi at the time. Here we go:

Here's the transcript of our conversation:
RVB: 00:03.336 Hello, everyone. My name is Rik. Rik Van Bruggen from Neo Technology. And here I am recording a podcast that is way, way, way overdue [laughter]. This is a conversation that we've been planning for a couple of weeks now. And on the other line-- other side of this Skype call, I've got someone that I've known for quite some time now, Sébastien Heymann of Linkurious. Hi, Sébastien. 
SH: 00:26.343 Hi, Rik. 
RVB: 00:27.218 Hey, it's great to have you on this podcast. Thank you for coming online. So, Sébastien, we've known each other, I think it's four years at least or something like that. And dating back to when you were still working for Gephi or you were involved in the Gephi project, I think. But maybe you might want to introduce yourself a little bit to our listeners. 
SH: 00:52.284 Yeah, it's been quite of a travel since then. So my name is Sébastien Heymann. I am currently the CEO of a startup called Linkurious. At Linkurious we provide graph visualization software on top of Neo4j. And we've started the company now four years ago at a time when we met the Neo4j people to better study how we could help with the graph visualization into the enterprise world. So I come from more of a scientific background. I've done a PhD in Computer Science and Complex Systems in Paris, in France, to study the dynamics of complex networks. And I have also contributed to an open-source project called Gephi. And Gephi had a pretty huge audience worldwide, with more than a million of downloads, thousands and thousands of research papers that cite Gephi for their studies. Gephi was, at a time, a software to help researchers extract meaningful information about networks. And we created an online community so that anybody could start to understand their data that they had. At that time, data was mostly stored in graph files. People exchanged graph files. And so the software was great to load a graph file and explore it completely. It didn't work when we started to use it in the enterprise world, outside of very advanced research in the development department. And we created Linkurious to better help analysts, business people, people who want to find information in complex situations to better understand it. 
RVB: 03:09.317 Yeah. Wow. Well, Gephi was one of the first things that I used when I started to work for Neo4j as a visualization tool, but also as a data exploration and loading tool even. It was very, very popular at the time. And is it still out there, isn't it or is it-- I don't know-- 
SH: 03:27.074 Yes. 
RVB: 03:27.982 Yeah. 
SH: 03:29.260 Yes, there is still an active community contributing on it, making updates. It's great-- 
RVB: 03:35.209 It's fat client software, right? It's fat client software. 
SH: 03:39.912 It is what? 
RVB: 03:40.840 Fat client software. You install it on your desktop and I think-- 
SH: 03:45.276 Exactly. It's a desktop-based software, so mostly for individual people. 
RVB: 03:51.811 And Linkurious is like the web version of that but maybe a little bit more than that, or how should I position Linkurious? 
SH: 04:00.643 I wouldn't say so because Gephi had a strong influence on Linkurious, but not in the way you could imagine. With Linkurious, we decided to create something which is really different in the way people experience graphs [inaudible]. In Gephi, it's for scientists, so you load everything even if you have a graph of 1 million nodes, you create a map of it. And you start to investigate. It's very complex for most of people. And people say also a lot of wrong things because they're not really understand the way they can use the tool and interpret, analyze the information in it. So with Linkurious, we wanted to take a totally different approach, where we first consider that data is too large. So we won't see everything. We just start with an entry point. And from there, because, for instance, you are a fraud analyst who is looking for a suspicious activity on an account, you will explore the activity of this account, the relationship it has to other customers and other contextual data you have. And because you dynamically visualize only what you want to answer your question, it's much more efficient in a business context. 
SH: 05:33.850 So with Gephi, we've taken some of the approach about what could be good layouts, how it could be-- interaction with the network. But here we wanted to make it for enterprise. So it comes first with the experience of navigating the graph and also the context where most companies now use web-based technologies. It is much easier to deploy. And we've created, with Linkurious, a sort of middleware on top of Neo4j, that quickly helps to deploy a business intelligence solution that is suitable for graphic databases. 
RVB: 06:18.622 I'm a big fan, as you know. So I've used it, I've seen it, I've demonstrated it at many, many different occasions. But what I kind of find interesting is that you mentioned that you sort of rolled into this from the scientific point of view. But what was it that attracted you, in the first place, as you personally, Sébastien? Why did you get into this [laughter]? 
SH: 06:42.699 It start to be a long time now, about 10 years ago, I will say, when I was a student at the French engineering university, Université de Technologie de Compiègne, in France, where I've had this crazy professor, a very good friend of mine even today, that was fond of networks, and especially web networks. How people would use the web to create information, to share it. The professor didn't have a technical background, but he was with more familiar territory. But he was very fond of some things that was emerging at the time called network science. And network science, which is a study of graphs, really pushed forward the ability to talk about complexity, to talk about networks, and to be able after that, to create tools. So, at that time, I fell in love with the ability to explore such complexity and to make beautiful maps, basically. 
RVB: 08:01.952 Wow. Was it one specific network, or one specific map that you thought was inspiring at the time? 
SH: 08:10.447 I would remember a map about web online communities, migrant communities, in particular, because we also closely worked with a laboratory of sociology, who studies how migrant group of people stay connected through the web, through forums, websites, blogs. And we created maps to better understand how they connect, how they keep in touch. And also to study what is the position of the institutional websites, for instance, what is the dynamic on the blogosphere at that time. And it was very beautiful studies. 
RVB: 09:06.599 Fantastic. Yeah. I mean, I think everyone has a favorite graph [laughter]. At least in our community though. I, personally, I once-- one of my favorite ones was a really small graph which was a social network of dolphins [laughter], which I thought was super interesting. Super cool. Well, Sébastien, you've been at this for quite some time, like you said. So what if I gave you a crystal ball and asked you to look into the future [laughter]? What would you see? What you see for our industry, for the science, for Linkurious as well, potentially? What's in the crystal ball? 
SH: 09:47.674 Well, for me, it's pretty clear what will happen for graphs. Because we're able to discover, not only on the scientific side, but also on the enterprise, thanks to the work of Neo4j to popularize graph database, that graphs can help solve-- answer complex questions. Even for machines to be more efficient, to quickly query complex context, but also to help humans better understand what is going on. So we still use [cases?] in fraud detection, T-money laundering, network management, and knowledge management in general. There is still a lot of new avenues where graphs will be useful. So I see graphs as a very important subset of data intelligence in general. And it is here to answer complex questions that statistics cannot answer it alone. And because of this value, it will become more and more important tomorrow to build complex solution that will involve graphs. So graphs will have a bright future and they are here to stay. It's been only 15 years from the beginning of network sciences. And we've seen a lot of inspiration from-- that came from science to create new businesses. 
SH: 11:41.405 For instance, the creation of LinkedIn, was inspired by the work of sociologists to study the six degrees of separation. Or Google, the famous [inaudible] algorithms, was also an inspiration from how we could create ranking in a network. And it happened that also from a scientific point of view first. And only with these two example, it created a huge diffusion, a very huge impact in terms of innovation and businesses. And we are barely get started with what could come next. 
RVB: 12:28.681 And I'm assuming that Linkurious also has very bright future ahead. And what's around the corner for you guys? Where is Linkurious going or what's in the new versions and stuff like that? What's in the future for you guys? 
SH: 12:42.918 Yeah. For us, it's a-- I will say it's a rich man problem. Because there are so many possibilities to apply networks that we grow also very quickly from what we are. And what we want is to create a company that will last and become the leader in what we call graph intelligence. In many situations, when we have to do business intelligence with graphs, Linkurious is the most easier tool to set up, to help experiments and to implementing [prediction?] the tool the analyst needs. So today, we are able to expand very quickly on our ability to provide our solution and the idea is to provide some things that can work in various contexts. 
RVB: 13:53.505 So lots more used cases, a lot more capabilities into the product. 
SH: 13:58.279 Exactly. 
RVB: 13:59.092 Fantastic. Well, Sébastien, thank you so much for coming online. As you probably know, we want to keep these podcasts fairly short. And we'll put a bunch of links to your website to some of the examples that I know you guys are really good at showcasing into the transcription. But I want to thank you for coming online. It's been great talking to you and I'll see you at the next graph event. 
SH: 14:25.287 Thank you very much, Rik. Also I would like to add a last thing. 
RVB: 14:31.819 Sure. 
SH: 14:31.910 Is a thank you for the whole team at Neo4j because when we started, we had no clue about who would like to use graph visualization in enterprise. So we've contacted you guys and you were open enough to provide us some information, and so that we could get started, and also continue provide more value to your customers. 
RVB: 15:01.169 That's fantastic. 
SH: 15:01.492 So it's helped a lot from the beginning. 
RVB: 15:04.454 And I think the feeling is mutual. I think we have a really good technology partnership and we're going to expand on it. So thank you, Sébastien. I'll talk to you later, okay? 
SH: 15:14.320 Thank you. Bye. 
RVB: 15:15.105 Bye.
Subscribing to the podcast is easy: just add the rss feed or add us in iTunes! Hope you'll enjoy it!

All the best


No comments:

Post a Comment