This topic came up in conversation elsewhere, inspiring me to do an annotated post of how I use Schema.org information to partially control how my book's metadata is presented on the Internet.
Intelligence for search engines
In an ideal universe, search engines would understand the context of the data that they retrieve. They would just know that a recipe is a recipe, that a book is a book, that a business location is a business location, and so forth. To the degree that they have gotten as far as they have, it's because of metadata — data about the data that they retrieve — that allows them some intelligence about presenting the information that they find.
To do this requires a combination of descriptive metadata from the data owner, and collation and presentation work from the search engine presenter. As in most such things, Google seems to be leading the way.
Google's Information Cards
When you search on a restaurant using Google, you get not only ranked links scrolling down the screen — you also get a nicely formatted “information card” on the top right of the screen that collects the information you would find most useful in an intelligent way.