Some textual analysis in my lost time about Blackbasta chat leak

[Axle-Bucamp]

Hello everyone,
Recently, for personal reasons, as a data scientist, I've been learning some tips and tricks about cybersecurity from leading YouTubers who are at the forefront of the field.

At one point, one of them discussed the Blackbasta chat leaks (and I might be a bit late to this), and since I was already working on a persona identification process to distinguish between bot speakers and humans within conversations, without being overly intrusive, for personal reasons, I figured I'd do some analysis.

Please don't judge too harshly any flaws in the graphs or things that could be improved, like connecting to other leaks, targeting specific locations, etc. I put this together in a day, and time's been scarce lately.

So here's my contribution, in case it helps anyone:
https://github.com/Axle-Bucamp/blackbasta-analysis
https://github.com/Axle-Bucamp/persona-detector

My takeaway? Major ransomware groups behave in remarkably similar ways, their "customer support" is faster than my Amazon deliveries, and certain keywords appear consistent across different groups.
If anything I've done here can help you solve problems, just reach out.

(The persona detector could definitely use some enhancements, better embedding models, improved cluster representation, refined dimensionality reduction... moving closer to something like Nomic AI's Atlas process, and so on.)