How does the Valence engine work?

Firstly, I know there are some threads on this, but they are all 5 years old.

I am looking to understand what information does Valence use and then what process it follows to do the recommendations. I will say I like what it does as I have often been surprised by the choices and added them to my favorites.

Thx!

@miguelito I’ve not come across a technical overview - only the posts that you’ve probably already read that describe Valence’s behavior from a user’s perspective. I’ll share what I’ve bookmarked in the hopes there’s new or useful information:

  1. Valence feature overview
  2. Brian Luczkiewicz’s brainstorming
  3. Brian Luczkiewicz’s rationale on not documenting Valence’s inner workings
  4. Discussion led by @armandhammer to understand Valence’s behavior

Ok very interesting. Thx for condensing the links.