Podcast Interview with Tom Zeppenfeldt, Ophileon

One of the topics in the Graph Database space is that is truly dear to my heart is the visual aspect. Graphs seem to be - for some very deep and profound reason, is my guess - a very natural way for humans to interact with data. And what better way to do that then in a truly visual way.

So I still remember 2+ years ago or something, this guy shows up at our Amsterdam meetup gathering and starts talking to me about the Neo4j browser interface - which was still in its infancy at the time. All of his questions made total sense, and I kindof wish I had had better answers for him at the time. But I kindof also am happy that I didn't, because that's why Tom Zeppenfeldt but on his "working gloves" and got to developing a new tool that looks really promising: Ophileon's Prologram.

Tom has recorded some really nice videos online where he is showing some of the capabilities of Prologram:

• how to work with Virtual Nodes in a graph
• how to enable Controlled Exploration of a graph

So that's when I asked him to come on the podcast too - and here's the result:





Here's the transcript of our conversation:

RVB: Hello everyone. Here we are again, recording another episode for our Neo4j Graph Database podcast. My name is Rik Van Bruggen. I work for Neo Technology, and on the other side of this Skype call, all the way in the Netherlands, is Tom Zeppenfeldt. Hi Tom.

TZ: Hi Rik. How are you doing?

RVB: I’m doing very well. How about yourself?

TZ: Yes. I'm very fine.

RVB: [chuckles] That's great. Well, Tom, most people who are listening to this Podcast probably don't know you yet, so would you mind introducing yourself? Who are you, what do you do, and what's your relationship to the wonderful world of graph databases?

TZ: Okay. Well, my background is not in IT. By education I'm an agricultural engineer and the main field where I've worked over the last couple of years was in international development in Africa and Latin America and that's also where there is a link between-- the link with Neo4j.

RVB: Oh, no way. Tell me about that.

TZ: Yes, as you probably know and there are lots and lots of types of development projects in agriculture, infrastructure, education and health for instance, and one of the typical things that you use in that context is what they call Result Chains, so diagrams that link activities and organizations and results and impacts and effects to each other and also a concept that is called Actor Constellation Mapping. That is a way of diagramming all the interests of different stakeholders in a project and how people collaborate, how they form networks, and  when I say networks I'm already talking graphs.

RVB: Yes, absolutely. Wow, that's great. How long ago did you first encounter Neo and how did that...?

TZ: That happened about four years ago and when we ran into a very concrete situation where we had to make these kinds of diagrams accessible over the Internet. So instead of drawing Powerpoints with graph-like structures in it, we were looking for a database platform that would allow us to share these things and make these things interactive over the Internet.

RVB: So, then that sort of created the desire to store them in a network way as well, I suppose.

TZ: Yes.

RVB: Yes. Okay. What are you doing now with Neo, Tom, because I know you're really doing some really fancy stuff, but tell us about that.

TZ: Yes. Well, currently our main project is a project that is partially privately funded and partially subsidized by the dutch government, and which is focusing researchers like journalist that do research or investigations on a specific subject. It include partially Neo4j for storing relations between documents, but also documents and content from social networks and also all kinds of tagging and metadata and for that, we have also a corporation with two universities in the Netherlands  that help us to automatically classify and tag documents or to detect relationships inside documents.

RVB: Wow, that sounds really great.

TZ: That's the main project we are working on now, yes.

RVB: Wow. As I understand there's also a big visualization component to that project, right. There's a lot of stuff that you're doing on how to bring that information visually to those researchers?

TZ: Yes, and indeed that's in terms of what we develop in terms of software is in fact, you can consider it an enhanced browser for the Neo4j database because once we were starting this project, we were running into-- well, we wanted to add more than the standard Neo4j browser, and that is why we now have a browser environment with multiple panels and panels that you can link to each other and in one panel you can display the data as a network, in other panels you can display data as a table and you can link them to each other, and that's what we call-- this product is called, “Prologram," for now, and that is what a number of developers are working on actually, yes.

RVB: That's very cool.  I've seen it live when you presented it at the meetup and everything, but I've also seen some of the YouTube videos and I'll post those with the podcast, as well.

TZ: Okay, great.

RVB: Maybe, just looking at it in a little bit more detail, what do you think is the most powerful aspect of this?  In other words, why did you end up choosing a graph database for doing this type of a project?  Any comments on that?

TZ: Yes. Well I come from … my background; I once had an IT company myself and that was typically relational databases, and finally we ended up building a metadata layer to mimic graphs on top of it.  In fact, that says it all, because the real world is far more complex than you can model, well in tables. Specifically the domains that I've worked in, be it agricultural development or now investigative journalism.

RVB: Yes.

TZ: If you have a very generic and basic structure like you have in Neo4j where also not just the notes but also the relationships are really things in itself, you call them first class citizens and are very descriptive. It allows you - even when your data model is changing or expanding and that is happens very frequently - You don't have to do a complete overhaul of what you already have to have and still have an optimize structure. So, you can keep on building and adding stuff and that's where for us one of the main advantages of Neo4j and that is on top of, of course the way that you can and really do optimized and very localized searches in your graph database, and I have the experience with building meta-structures on top of relational database then the number of joins is incredible and finally it makes it workable.

RVB: Yes. I don't know if you've listened to any of the other podcast episodes but some of the Neo4j founders have been on there as well and that's how they started Neo4j. They started developing a meta-layer or graph layer on top of-- it was Postgress at the time I believe. I think you're coming to do right conclusion there [chuckles].

TZ: When the...

RVB: Go for it. You wanted to add something?

TZ: Well, for us it was quite easy to pick the concepts, to understand the concepts and that's also I think we now are making nice progress with what we make as a generic browser on top of Neo4j is that we come already,  we came already from a graph, a mindset that knew the advantages of a graph database.

RVB: So what is the future of all this? Tom, where do you think this industry is going, where should Neo4j go? Where is the Prologram going?

TZ: Well, that's the-- You said it's ten minutes, this podcast. [laughter] [crosstalk] RVB: Well, let's limit-- Okay, well, I think graph database are there to stay, it's not the hype. Especially when relations between contents become more and more important than perhaps the properties of content… For instance, we are now working on recommendation engines and while, in the the beginning, people got recommendations on the basis of properties or links inside the content itself, so very explicitly - now  we are already doing test with integrating into recommendation, the social aspects. Then you finally end up concluding that the suggestions that you can make on the basis of how people use the contents and that is typically something that you can easily store in the graph. They are better than the suggestions that you can make on the basis of the explicit tagging of a content.

TZ: As the world gets more and more connected, whether it's Internet of Things or documents that are shared. Well, it's the graph model is a very nice way to describe - you get more exact models than a relational database. I think the position of graph databases in the world is an established one. What we are now doing with the Prologram platform, we have been building it with let's say, about three to four persons of the last 14 to 15 months. We now have a version that is -- on one end it's good enough, but we are; say in the next two to three months we will probably start sharing it with the world because we want to know how people interact with it, and also I'm sure you cannot keep on developing without user feedback. We have already quite some users and in terms of functionality there will be additions like well integration with the elastic search, a complete function and trigger system that allows you to change theories, virtualization of notes that allows you to make aggregations - to visualize aggregation of notes and relationships. So, that's where we are more or less going for the next quarter. That's more or less the road map that we have.

RVB: You discover it as you go. Right?

TZ: Yes.

RVB: It sounds really exciting. We'll wrap it up at that. Tom, it was really great talking to you. Thank you so much for coming online and apologies for the technical hiccups that was entirely my fault. But, thank you for coming online. I'm sure will see each other again at one of the next meet-ups.

TZ: Okay. Thanks for the opportunity and keep up the good work.

RVB: Thank you. Cheers, bye.

TZ: Okay. 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

Rik