At PASS Summit 2017, Shreya Verma and I will be co-presenting a session on using the newly added Graph data (nodes / edges) processing capabilities in your applications. One of the patterns we will discuss in that session is how to leverage in-database Python scripts to detect “similar” nodes and thereby ‘infer’ edges in the graph.

Update November 4th, 2017: You can now download this graph as a PDF file: PASS 2017 Sessions. Please view with Adobe Acrobat Reader at zoom level 1600%; web browsers like Edge and Chrome do not allow you such high zoom levels.)

We have a very interesting graph processing demo which uses PASS 2017 schedule data. In this demo, we will start with each session as a node in our graph, and then use in-database Python to detect similar sessions (based on the session title, abstract and the speaker names.) We then return that data to SQL to build weighted edges in the graph, and then use an external visualization tool to draw the graph. As a sneak preview, here is a section of the graph built with this method; edges connect similar sessions, and the thicker the edge, the more similar those sessions are:

To know more on this theme of how Python and Graph integrate, and how we generate the above visual, come see us in our session on Nov 3 @ 3:30PM. Yes, we know that’s the last session slot of PASS Summit 2017 – you always save the best for the last 🙂