It's a large graph database that I've prepared myself over two years, as a kind of part-time hobby. It's largely an automated scrape of metadata available on wikipedia, goodreads, ISBNdb etc.

Intersections are sourced from public ratings and reviews across the web. E.g. these books are all liked by one person so are can be made to intersect in the db - https://radicalreads.com/noname-favorite-books/