The Weekly Edge: Read S3 Data with Cypher, Graph in Fabric GA, Epstein Files Graph Viz, & More

[News:] Custom Graphs in Microsoft Sentinel Graph Enter Public Preview

As part of RSAC 2026 happening this week, the Microsoft Sentinel team announced that Microsoft Sentinel Graph will support custom security graphs in public preview starting on April 1, 2026. 

Blog author Suresh Palani explains that custom graphs let you build security graphs tailored to your unique security scenarios using data from Sentinel data lake as well as non-Microsoft sources. With custom graphs, you can uncover hidden patterns and attack paths, and help surface risks that are hard to detect when cybersecurity data is analyzed in isolation. 

In case you missed it, Microsoft Sentinel Graph was announced in public preview last October (previously covered by the Weekly Edge) and then Sentinel Graph hit GA two months later. It was Microsoft’s John Lambert who once said, “Attackers think in graphs. Defenders think in lists,” but now with Sentinel Graph, defenders can think in graphs too.

[Repo + Paper:] Generate Synthetic RDF Knowledge Graphs + OWL Ontologies with PyGraft-gen

RDF users and developers will appreciate a big open source repo (+ a research paper preprint) from the Orange and Zalando teams that dropped earlier this week.

PyGraft-gen uses stochastic generation to produce synthetic RDFS/OWL ontologies and RDF knowledge graphs with reliable structure while respecting OWL constraints, making it ideal for testing AI pipelines, benchmarking graph algorithms, and research scenarios where real data is sensitive or unavailable.

Project contributors and co-authors Nicolas Hubert, Lionel Tailhardat, and Ovidiu Pascal knew the challenge of generating synthetic RDF knowledge graphs that are simultaneously ontology-compliant and structurally realistic. Comparing two previous generation paradigms, they created PyGraft-gen (an updated fork of PyGraft) and tested its effectiveness via three subgraph matching techniques – SPARQL, SHACL, and VF2++ – to verify that the generated knowledge graphs reflected realistic domain patterns. 

Whether you’re hands-on or more theoretical, there’s lots to dig into with this one.

[ICYMI:] Explore the Epstein Files as a Graph Visualization

The Epstein Files have been all over the media the past few months, and now you can explore the trove of publicly available data about Jeffrey Epstein and his human trafficking network using a network graph visualization. Originally launched in February 2026, the Epstein Exposed database was recently covered by WIRED magazine and continues to add more information on a regular basis.

The comprehensive, publicly accessible database of the Jeffrey Epstein case files indexes over 2,146,580 documents, 1,603 persons of interest, 3,615 flight records, and 1,783,792 emails from U.S. court records, DOJ releases, FBI disclosures, and congressional investigations. The site features interactive tools including a network graph, flight map, forensic calendar, email viewer, research chat, document review queue, and cross-reference search.

Created by indy data engineer EricKeller2, Epstein Exposed is free, open source, and entirely community-funded. The project is not affiliated with any government agency, law firm, media organization, or political party. It’s also a perfect graph viz use case.