Why interest in graph databases and graph analytics are growing

The Data Exchange Podcast: Paco Nathan on recent trends in large-scale graph technologies and novel applications of graph databases.


SubscribeApple • Android • Spotify • Stitcher • Google • RSS.

In this episode of the Data Exchange, our special correspondent and editor Jenn Webb organized a mini-panel composed of myself and Paco Nathan, author, teacher, and founder of Derwen.ai, a boutique consulting firm specializing in Data, machine learning, and AI.  We began our discussion with a brief tour through the highlights of the upcoming 2021 NLP Industry Survey results.  The free report containing the survey results will come out in a few weeks, subscribe to our newsletter and we’ll notify you when the report comes out.

Of late, Paco has been doing a lot of work with graphs and as such he’s had to immerse himself in the world of graph data management technologies. This conversation is focused on what’s new with graph databases, and why there’s been a resurgence in interest in them. We also discuss use cases of graph databases, graph analytics, and graph neural networks. If you’re an engineer who digs graphs, note that Paco and Derwen are hiring

Paco Nathan:

When you’re working with graphs, one thing is that when you see problems like complex disambiguation problems, or, you know, a classic example would be like transitive closure. If you try to run those in an SQL engine, you’ll bring it to its knees. SQL is built for being able to index with normalized relations, tightly controlled relations, across different indexes and tables. And if you take arbitrary relations within the same table and join it to itself, many, many times, which is what you end up having to do with graph technologies, your typical signal engine will just come crashing down.

… There are about 30 vendors in the graph database space, and they tend to follow along a couple lines. One is they’re doing W3C Semantic Web kind of stuff, where you can do a lot of really interesting inference. And there are reasons, but they tend to fall back toward this old school thing called the triple store. I have no idea why people are still doing this in the 2020s. And then there’s something called label property graph, which relaxes some of the constraints and allows for really much better information management. But it doesn’t really follow standards yet. And then there’s other stuff that’s just crazy

Subscribe to our Newsletter:
We have two free reports coming out in a few weeks: 2021 NLP Industry Survey Results and the 2021 Data Engineering Survey Results. Subscribe to our newsletter and we’ll let you know when the reports are available.

Related content:


[Image: Play with UV light by Pietro Jeng, via Unsplash]