No sure if we should be happy or sad - but hey - the Covid-19 summer of 2020 is almost behind us. Like most people, I found it quite a strange and unusual summer, with very few foreign adventures (although I did manage to squeeze in a cycling/camping trip to the French Alps in July), lots of cycling, some great family time... and of course lots of time with graphs :) ...
So that means that we are also kicking the Graphistania podcast back into gear - here's the next episode for you:
Here's the transcript of our conversation:
RVB: 00:00:15.863 [music] Hey, Stefan, I do need to ask you for consent, I think, right?
SW: 00:00:19.847 Hi. Yeah, I consent. [laughter] This is always the weird moment.
RVB: 00:00:24.727 Exactly. I thought, "Start with that one again."
SW: 00:00:28.436 Exactly, just to create a little bit of tension in the air.
RVB: 00:00:31.537 Indeed. Hello, everyone. My name is Rik, Rik van Bruggen from Neo4j. And here we are, again, after the summertime. The autumn is almost upon us. We're going to enjoy a couple more weeks of better weather, but we wouldn't be able to do that without another podcast episode. So on the other side of this Zoom call, I've got my dear friend Stefan, Stefan Wendin. Hey, Stefan.
SW: 00:00:58.170 Hi. Yeah, Autumn is coming upon us. That's for sure. Here in Sweden, it starts to get cold. But let's not talk about the weather; that's just going to be depressing. So let's stay focused, happy, and on all the cool things that has happened.
RVB: 00:01:13.091 Agreed. And one of the cool things, I guess, was summer. How [was your?]--
SW: 00:01:17.542 Yeah. [crosstalk] memories. [laughter] Don't ask me about anything now. It was good. Yeah. I think from--
RVB: 00:01:24.397 [crosstalk]--
SW: 00:01:25.233 Yeah? You want to go?
RVB: 00:01:26.792 I just wanted to ask you how did your summer go? And, professionally, of course, how has the world been evolving for you?
SW: 00:01:38.711 Yeah. As for the summer part, I usually spend a lot of time reading, studying, [inaudible], and of course, always checking out all the cool parts in the graph world. Strangely enough, this year, I really didn't get into that kind of calm, relaxed mode, or at least, it was a little bit harder, and I think, mainly, because of this thing that people call the new normal, which is not my favourite word for sure. But anyway, things has evolved. Nowadays, I've been, instead of travelling, working mostly from my remote office alone and so on. So I think that kind of changed, that part. So seeing the list of great ideas or topics for this call just get me all energised again. So feeling a little bit behind on that part, but I think that's also healthy to just admit that we can't be perfect all the time, so that's good.
SW: 00:02:36.611 As for the professional life, I think a big part of what we do has changed completely. I mean, what I do is running the innovation lab here at Neo. A big part of that is, and has been, a very hands-on person-to-person experience, like going into a [inaudible] room for a week and then just crunching away, taking care of all of these things, and obviously, there is no travel and there is also restrictions on rooms. So what I did was - as I always do when I hate something - because I really hate bad Zoom meetings and boring things, right, the story why I start the lecture at [inaudible] is because I hated sitting there listening to boring shit. So what I did was look upon this and see how can I transform this into a remote offering. Because I think in this time, business will not stop, right? But more than ever, it will be more important to think about the validation, and basically, that's what we do. Basically, this kind of scientific methods of seeing if graph is a fit for you and doing that from, basically, a technology and data point, of course, but also bringing in the business and the use of perspective in the same kind of loop. So we put that together and have been delivering pretty much nonstop before summer, and now it's, again, nonstop going.
SW: 00:04:03.823 So it's a super popular product, which I think is kind of cool. And, actually, it is better remote, which is strange, in a sense, because it's against everything I believe. I believe, of course, in the meetings of people, and I had a very hard time accepting that this becomes so good. It was kind of a shock to me. But I think one of the cool parts is that we get to kind of work with people from all over our company, which is super cool. So picking the brains of our data science team and people from New York, for example, and bringing them into our European customers without so much trouble of travels and so on. So I think that brings in [enterprise?]. And I think, again, this validation part comes related to what I see. A lot of the innovation stuff, usually, is the crazy persons in the company, right, the one with the stupid ideas. And in a time of scarcity or a pandemic or where we are, those projects tend to go on hold, right? So I think, more than ever, it's important, actually, to do this validate thing, to be sure that if I can validate my hypothesis faster than my competition, I'm going to stay afloat, right? So I think this is, basically, how it helped me selling this in the new world and adding my perspective to that. So it's very helpful in that sense. Do you see anything on the selling part, Rik, from your side?
RVB: 00:05:32.461 Yeah. I guess so. There's quite a bit of changes, right? And just like you, I've had to take my job into a new direction. Selling, usually, is all about connecting to people and talking to people and just making them see the value of the products that we offer and the services that we offer, right? So it's very different. And I feel like the biggest difference that I've kind of experienced is that before the pandemic hit, I think lots of people were still experimenting a lot more. And they actually had budgets for that and they were able to do trial and error a little bit more and those type of things. Whereas today, I find that the importance of a solid business case, a solid value case for these type of projects is even more important. Well, it's always been important, but it's only become more important. And the thing that I constantly try to help clients with is to make those cases be stronger because I mean, in most cases, if people are looking at a fraud detection system these days, they kind of know that the individual transaction isn't going to be fraudulent anymore. The fraudsters are smarter than that. You know what I mean? They know that the fraud is going to be in the pattern, in the connection of transactions.
RVB: 00:07:07.128 And so as a consequence, I need to be able to justify much more explicitly with the client, "You know what, if you're going to combat fraud in 2020, you'd better have some kind of understanding and insight into the connections between your transactions." So there's a lot more emphasis on that and a lot more work that we do with clients on that type of stuff. So yeah, it really kind of has transformed my job, but it's also enhanced it in many ways. I think I find that it's not the worse situation, I wouldn't say. Obviously, I would love to be visiting clients more and talking to real people instead of the umpteenth Zoom call, but I do find that it's a healthy thing to talk about technology and justifying technology a little bit more explicitly than we possibly have been doing before. So I'm kind of happy about that. It's a good evolution. Let's hope, from this pandemic, we can remember the good things. You know what I mean? That's something that I'm hoping for.
SW: 00:08:24.311 Yeah. [crosstalk]--
RVB: 00:08:24.481 So one thing that we should be doing, Stefan, is we should talk a little bit about all of the cool things that our users and our community and our customers have been doing in the past couple of months, right?
SW: 00:08:35.424 Oh, yes.
RVB: 00:08:36.851 Did you see anything nice pop up in the past two months that you want to talk about?
SW: 00:08:42.037 Yeah. I mean, first of all, when opening the list, I think this is one of those where we should just share the entire list of good stuff that we looked upon.
RVB: 00:08:50.488 Of course.
SW: 00:08:50.811 But one of the things that really blowed my mind was this whole idea of graph embeddings. We have been talking about it for a little while, but finally see it coming in the release, it's like, "Holy [crap?]," to just speak it out loud, and how it kind of allows me-- sorry for that, again, this is just who I am. [laughter] I can't change.
RVB: 00:09:13.774 We forgive you then.
SW: 00:09:14.212 I'm too old to change. No, but this idea kind of helped me represent or embed my graph and represent it as vectors or sets of vectors, meaning that I can take a single node and embed it or I can take a path and embed it or I can take the entire graph and embed it. I think it's so fundamentally different because then I can feed it into my machine learning [inaudible], enrich it, and then learn and feed it back and to kind of create this kind of loop back and forth, almost. And for me, this is one way of adding context and taking the best from two worlds and combining them. And I think it's one of those moments. Usually, there is a couple of these in your career. One where, very early on for me, playing music, when I saw a sampler the first time, an instrument with no sound. And you can record whatever sound you want. It's literally insane. The idea's like, "What?" And then I know, with Photoshop, it was the same with the cloning stamp. [laughter] Again, taking something and painting with that thing that you sampled from somewhere. It's like this is literally the same thing but within machine learning and graphs, right? Doing that, which seemed almost like you can understand it as a concept, but then seeing it happening, it's super good. So I think, for me, this one thing that comes from the remote way, to be frank. I was on a call. I brought in Alicia from our team with one of our clients. And then this week, I had-- of course, they are joining both from New York, so this is one of the possibilities that just this strange world has opened up. So supersizing like that.
RVB: 00:10:54.905 Amazing. Yeah, it's really cool. And myself, as well, I've been working in the graph industry for quite some time, but this was kind of new to me, and I think it's extremely powerful. And for those of you that don't know this-- and you'll see it in the transcription. We'll put some links there. But Neo4j has this graph data science library these days that now supports these graph embeddings that you can then use and reuse in machine learning. But we'll put some links on that. Can I suggest maybe something that happened to me this summer, and I can talk about it?
SW: 00:11:32.838 Yes. I was just thinking about it. Yeah, what have you been up to?
RVB: 00:11:36.394 Yeah. Well, I mean, there's a couple of things that I'd like to highlight, right? But there was this one beautiful experience in July. I've actually blogged about it as well because this is the local newspaper here in Belgium. It's the financial newspaper, like the Financial Times or the Wall Street Journal of Belgium, if there's any such thing. A couple of years ago, they had done this publication about the executives of public companies, Belgian public companies, right? And it's a very short list. There's like only 150 listed companies, so there's only - I don't know - a couple of hundred executives, right? And I want to say like three or four years ago, they published this article. And I was like, "Wow. There's graphs on this article." I go digging for it, and I find the data set embedded in some Javascript. And I wrote about it at the time because I thought it was an interesting case to kind of understand the networks between the executives of the public companies, right? So now, this summer, in July, all of a sudden, I get a phone call from a journalist from that newspaper saying, "Hey, Rik, remember when we did that thing a couple of years ago, and you basically stole our data?" And I was like, "Yes. Yes, I do remember." [laughter]
SW: 00:13:06.761 Uh-oh. Am I in trouble?
RVB: 00:13:07.680 Am I in trouble? No. And the guy goes, "No, no, no, no. That was great. That was great. And we want you to work on the new version with us." So he basically gave me the updated version of the data set. We ran a bunch of algorithms on it, and we basically wrote a new article for the newspaper based on that analysis, which was super cool to do, and I've blogged about it. And it was actually super interesting because it kind of highlighted a couple of really important new players in that old boys' network, and the most interesting part of it was that the big fish in the old boys' network was a girl. [laughter]
SW: 00:13:54.192 Yeah. That's very interesting, exactly what I was saying. I'm looking at the visualisation, and I can just imagine that it's also representative of almost [inaudible]. I think you run it also.
RVB: 00:14:05.878 Yes. Yeah.
SW: 00:14:06.136 And you see that that big network, there's a female in the middle there. Ooh, this is interesting, just like my family.
RVB: 00:14:12.643 Yeah. That was a really cool thing to do. And yeah, I really enjoyed that. And then there was another article that popped up end of July around the Proteomes of Life. I don't know if you've read that.
SW: 00:14:24.963 Oh, yeah. Yeah, I should look into it. It was one of those you click on it, and it just unfolds and unfolds and unfolds and unfolds, and you never know when to stop. So I'm in the midst of that Pandora's box thing. But I think it was super cool. It reminded me also about one of the books I read this summer, which is called the Blueprint of Robert Plomin. It's about this idea of genes in a sense, or basically, nature versus nurture, right? So environment versus the genetic construction of you and how that create situations and so on. So a great book to read if you haven't read it. A little bit controversial, so try to read it with - what is it that we say here at Neo? - positive impact, right?
RVB: 00:15:15.160 Okay. Very cool. Yeah. Well, I thought that article was also--
SW: 00:15:17.698 Yeah. But you have a greater story.
RVB: 00:15:20.707 Well, that article was really interesting for me because it reminded me of one of the earliest Neo4j experiences that I had with the University of Ghent here in Belgium. They were doing what they call [metre?] proteomics, which is the analysis of how proteins interact with one another. And they had built this tool to visualise and analyse the interactions between proteins. And it was a really fun experience. We even organised a workshop around it at the university and all those types of things. But what I still remember very fondly about that work that we did at the time was that the first experiment that they did to highlight the interactions between proteins was actually around beer brewing. [laughter] So--
SW: 00:16:13.394 Exactly. So for all those listeners, this is Rik van Bruggen, famous for the beer graph. [laughter]
RVB: 00:16:20.744 Exactly. So it was meant to be. That was all about how different yeasts in a beer would interact with one another and how they would influence process optimisations. If you brew a beer at one temperature or at a different temperature, obviously, the yeasts and the proteins of those yeasts will react differently, and you will have a different flavour and a different beer as a result. So it was a super cool case, and the proteins of life kind of reminded me of that, so.
SW: 00:16:54.957 Yeah. No, but I think, yeah, it's very, very interesting. And it is one, again, of these kind of things that we say, sometimes, as a tagline. Graphs are everywhere. But they really are, in a sense, right? And I think this whole idea of this great invention of the box which then became the table is, sometimes, good. But it's also just so obvious that, of course, there is dependency sort of things, which is super complex and has a network structure in everything we do. Also, this one, it made me remember, we talked about folding way back, one way of my favourite [ones?].
RVB: 00:17:31.088 We did.
SW: 00:17:31.991 Protein foldings. Again, a fantastic use case on that as well. But cool.
RVB: 00:17:37.905 Very cool.
SW: 00:17:39.372 Did you have any other good ones there?
RVB: 00:17:41.084 Well, I mean, there's a zillion other ones. There was the Summer of Nodes, which I really enjoyed following - that was our colleague Lju that organised that - with a bunch of really interesting articles. There was CompoundDB4j, which I thought was an interesting and very complicated name for creating these drug resource databases, really interesting. And the one that I kind of picked up on - it's also my not so hidden history as a software geek - was the code analysis bit. So there's quite a few people that are using Neo for analysing dependencies, not between regular things but between code, right? That code refers to other code and stuff like that. And there's actually a tool that our open-source community has built, something called jQAssistant, a really, really cool tool that allows you to do all that stuff semi-automatically, which is pretty amazing. So I enjoyed reading that as well. That was a cool article as well, so.
SW: 00:18:57.770 Yeah. I missed reading it, to be fair, but I know we talked about it a little bit, and it sounds very, very interesting. So I'm going to give it a try even if it's a little bit out of my turf. I think this is also how we grow. If it's not a little bit uncomfortable, usually, you're still in the same old same old, right? So I'm going to give it a try. And I suggest all of the people thinking like, "Ooh, this is too technical for me," just do it. And this is how we grow as humans, so yeah. I'm going to just--
RVB: 00:19:28.038 Very true.
SW: 00:19:29.001 --role model that behaviour.
RVB: 00:19:30.988 Grow as humans. That sounds like a fantastic ending though, doesn't it? [laughter] So--
SW: 00:19:34.966 Yeah. So now we've got to just shut up and end it.
RVB: 00:19:38.174 Exactly. Hey, Stefan, I really enjoyed talking again. We'll do some more podcasting over the next couple of months, right?
SW: 00:19:45.848 Yeah. That's for sure. And again, if you have any requests that are things we should pick up and talk about, feel free to send it to us. We like any sort of feedback, or I like it, and then Rik likes it--
RVB: 00:19:59.180 Me too.
SW: 00:19:59.792 --because we're going to laugh about it together.
RVB: 00:20:01.778 Exactly.
SW: 00:20:02.422 So perfect. Not a [inaudible] task then.
RVB: 00:20:04.394 Cool.
SW: 00:20:05.400 Bye-bye.
RVB: 00:20:06.187 Thank you for your time. Talk to you soon. [music]
All the best
Rik
No comments:
Post a Comment