Monday, 26 September 2016

Podcast Interview with Sascha Peukert, TU Dresden

Another week another podcast episode! Here's a great chat with a graphista that has not even started his professional career yet - but that, by the sounds of it - could really be a stellar one. My chat with Sascha Peukert in Germany:

Here's the transcript of our conversation:
RVB: 00:04.386 Hello, everyone. My name is Rik, Rik Van Bruggen from Neo Technology, and here we are recording another podcast session for our Graphistania podcast. And today I have a very interesting guest on the podcast from Germany - from Dresden. And that's Herr Sascha Peukert. Hi, Sascha. 

SP: 00:21.410 Hi. 
RVB: 00:21.774 Hi Sascha, I know you're a master's student at the Technische Universit├Ąt Dresden
SP: 00:27.609 Yeah, that's right. 
RVB: 00:28.531 [chuckles] I try to pronounce it correctly. But would you mind introducing yourself a little bit to our listeners, please? 
SP: 00:35.595 Sure. Yeah, I'm a student from Dresden currently finishing up my studies - Master of Computer Science. I mainly focus around social engineering and database systems, of course, and currently writing my thesis about Neo4j
RVB: 01:01.116 Wow, then I think you're-- one of our dear friends who actually started this podcast with me, Michael Hunger is part of that work, right?
SP: 01:10.321 Yeah, he's one of my advisers. 
RVB: 01:13.268 Oh super. Michael is a dear friend. And what's your thesis about, Sascha? Can you tell us a little bit about that? 
SP: 01:20.855 I write about graph views, using the example of Neo4j. So if you have a relational database systems, you have this functionality we called views. Views are just queries stored in the database which can be called upon or reused by other queries. And what's cool about that is you can configure access to those views, and also to the tables themselves. So, you can permit one's access to the table, but you can allow access to, for example, a view that only shows certain parts of the table or join tables. And you don't have that kind of functionality in Neo4j right now, but I'm working on establishing that. 
RVB: 02:17.812 You're actually implementing some of that? 
SP: 02:19.275 Yeah. 
RVB: 02:21.749 Wow. I'm looking forward to that [laughter]. And you're supposed to graduate this summer, right? 
SP: 02:28.990 No, not really. 
RVB: 02:30.097 Okay [chuckles]. 
SP: 02:32.605 It would be around February. I just started on that. 
RVB: 02:37.307 Okay, well, good luck. So, how did you get into graphs, Sascha? Can you tell us a little bit about that, and then what's your interest in it, and why do you like them? 
SP: 02:48.855 The first time I came in contact with graphs, other than our first semester math course, was in December of 2013. I was attending a meeting of the Java User Group Saxony, and I heard Dirk Mahler talk about his new graph based code analysis tool called jQuery. 
RVB: 03:13.435 jQAssistant. 
SP: 03:14.732 jQAssistant. 
RVB: 03:15.231 Yes, exactly. Yeah, yeah. I've interviewed Dirk for the podcast as well [chuckles]. 
SP: 03:19.964 Yeah, I've heard that. And after that, I was really was hooked on Neo4j and Cypher in particular. It was just really cool to see that you can emphasize the relationships about data and not have to deal with all this join problematic that relational database systems come with. 
RVB: 03:53.411 And what have you used it for? I mean, what is the type of work that you've been using it for? I've read your Wikipedia [chuckles] graph gist, which is very similar to something I did, right? 
SP: 04:05.421 Yeah [chuckles]. 
RVB: 04:08.603 What have you used it for? What have been some of your used cases? 
SP: 04:15.637 Well, just put little experience with my Wikipedia data set for example. I also did a student research project on parallel graph algorithms on Neo4j, looking-- 
RVB: 04:34.023 Tell us more about that? What was that? 
SP: 04:37.235 I tried to show how to add more code to the Neo4j  server in a way that you can call upon that, and also emphasized on how you can parallelize this code, so it will run more efficiently, quicker, all those things. But sadly, that was prior to Neo4j 3.0 so I don't have procedures available. 
RVB: 05:14.307 Well, maybe it's time to revisit it then, Sascha [laughter]. 
SP: 05:20.431 I'm thinking about starting a few blog posts about that project, because I sadly wrote in German and therefore nobody has ever read it [chuckles]. 
RVB: 05:30.993 Yeah, that's always a problem for me as well. I have to first force myself to write in English, because otherwise it just doesn't get read as much. But that parallelisation sounds really interesting. There's a lot of work that's been going on in the procedures ecosystem basically since last April, right? And a lot of interesting things going on there. So, what does the future hold for you, Sascha? Where do you see this going? What are the cool things on the horizon? 
SP: 06:08.379 That's a really good question. For me it's finishing my work. 
RVB: 06:14.151 So, no holidays. It's all work, right [chuckles]? 
SP: 06:15.762 No holidays, sadly. But about Neo and graphs, I guess this whole industry will just become much more important than it already is, and we will see-- I guess we will see lots of new, interesting use cases. With my GraphGist Wikipedia game, I try to distance myself from those traditional use cases, and I think we will see much more diverse things done with graphs, like the things about Panama papers
RVB: 07:07.143 Yeah. I think the Wikipedia challenge or game that you wrote about, it's a little bit of a silly game in a way but it's also-- 
SP: 07:16.814 Yes, yes, of course. 
RVB: 07:17.752 --but it's also really interesting because you can ask questions that would be extremely difficult to ask in any other database, right? If you tell someone, what is the connection between two things, how do you do that in SQL? It's almost impossible, right? You have to basically hypothesize on all the possible connections and then the database will say yes or no. But you can't really say what are the connections. It's actually a very interesting example that you find use cases for, in banking, in fraud, in whatever. It's actually very representative, I think. Sorry to do all the talking here, but-- [chuckles] 
SP: 08:03.440 No, no, you're absolutely right. It's so obvious if you start working with graphs that so much data is just a graph. 
RVB: 08:17.758 It's fascinating, right? It's one of those things that we should have a chat about it over a beer some day [laughter]. 
SP: 08:25.403 I'd like that. 
RVB: 08:26.201 Yes. 
RVB: 08:27.037 Well, Sascha, thank you so much for coming online. As you know I keep these podcasts fairly short, but we'll put some links to your work on the transcription page when we publish the podcast, and then hopefully people can enjoy or Google Translate some of your work [chuckles]. That will be interesting, I think. Thanks a lot, Sascha. 
SP: 08:48.529 Thanks for having me. 
RVB: 08:49.522 Have a nice day. 
SP: 08:51.051 You too.
Subscribing to the podcast is easy: just add the rss feed or add us in iTunes! Hope you'll enjoy it!

All the best

Rik

No comments:

Post a Comment