Ok, so the title might sound a bit confusing, but here's an analogy of what I'm trying to achieve. Let's imagine that we have the following dataset:
1.4m articles
1.4m replys
5 comments
56.9k users