The Graph Intelligence Stack

Leo Meyerovich on how and why the best companies are adopting Graph Visual Analytics, Graph AI, and Graph Neural Networks.

Subscribe: Apple • Android • Spotify • Stitcher • Google • AntennaPod • RSS.

Leo Meyerovich is founder and CEO of Graphistry¹, a startup building tools to democratize visual graph intelligence and graph machine learning. Leo and I recently wrote a well-received post (“What Is Graph Intelligence?”) making the case for why companies need to revisit graph analytics and graph intelligence. We believe that it’s just a matter of time until graphs become a standard method to analyze data for operational insights. In this episode we discuss the key insights from our post, and Leo provides updates on the rapidly changing ecosystem for graph intelligence.

Download our FREE Report: 2022 Trends in Data, Machine Learning, and AI

Leo Meyerovich:

If we just start on the visual side, we think of a few categories here. Some of the classic “connect the dots” scenarios. A lot of our work is forensic where there’s a security or fraud incident, or physical people interacting. Often there’s actually metadata in it. Nowadays everyone’s dealing with digital data. So for forensics it might be people are clicking around a website, and maybe they’re a bot ring trying to defraud your website. We also get into a lot of big companies where they don’t see their supply chain, they don’t know their HR hierarchy. And that’s a lot a lot of the connect the dots stuff.

The other side is more of a data science perspective where one way of thinking about an edge in a graph is a correlation. … And more broadly, If you’re looking at different correlations, and a lot of disparate data sets, you want to really get those fuzzy correlations. You’ll end up with a graph and having visual tools early in your data process … let’s you actually reason about those relationships. If you want to make machine learning interpretable, one of the things I’m seeing is while a graph is not the end all, it’s actually one of the more powerful tools for showing correlations.

In our recent post we highlighted the explosion in interest in graph neural networks (GNN) within the research community. As this begins translating to industry interest in building and using GNNs, companies will need tools for data processing. As we noted in our post, companies can use modern data platforms and tools like Spark to perform most Graph Intelligence tasks. But as Leo notes in this episode, some large-scale data processing requires graph computations that aren’t convenient to do with existing open source data processing tools. This is a gap in the suite of open source data processing tools that aspiring open source developers might want to fill.

Highlights in the video version: