Matt Jockers posted the slides of our recent talk, which include some explanation of how similarity is determined.

Well, that's disheartening