A simple example

I visit a library, and ask the librarian to suggest a book.
For a librarian suggest books to you, they need to understand both the library system, and what you are looking for.

What does the library look like?

Oversimplified, of course, but below is an image of a small section of a library. To find a specific book to recommend to me, the librarian sees a lot more than what we see.

What does the librarian know?

The librarian looks at the scene, and understands a great deal of context about the world of books! Not just everything about books, but where they live! Fantastic skills for finding just the right book for a visitor.

What is the librarian data model?

Much of the data from the previous drawing is not useful. So we can remove that. And now we want to add the relationship rules.

A great suggestions

Given this complex model, the librarian can say “I can easily suggest to you a book which is just behind the stool, on the bottom shelf.” Or better yet, the librarian can go get it and hand it to you. This is in essence what Hubstream does. Except we can do it automatically, with several million libraries.