At an event hosted at their California headquarters today, social networking giant Facebook announced Graph Search. It is Facebook’s “third pillar” following News Feed and Timeline which allows you to search everything shared with you by your friends.

Facebook was clear to draw a line between a typical web search and Graph Search. “You want a search tool that can help you get access to things people have shared with you,” said Facebook CEO Mark Zuckerberg. Graph Search uses your friends’ locations, likes, comments, photos, tags, and much more content to deliver results you’re looking for.

The social network makes clear that Graph Search takes all of a user’s privacy settings into account. In other words, unless you share material with someone on Facebook, they won’t be able to find it while searching.

Facebook is taking a natural-speaking approach to Graph Search. By typing in a phrase like “My friends who live in Palo Alto who like Game of Thrones,” Facebook will be able to know exactly the type of friends you are seeking. This specific example would provide a great way to figure out which local friends to invite over for a Game of Thrones party.

In another example presented at the keynote, a user searches for “My friends who like Star Wars and Harry Potter” and is only presented with friends who like both franchises. You can even do broad yet specific searches like “Photos of my friends taken in national parks.”

Graph Search will be placed at the top of Facebook and appears to take over the entirety of the existing blue Facebook menu bar. The service will initially focus on People, Photos, Interests, and Places, but Zuckerberg notes it’s in beta and will expand further.

Naturally, critics will be concerned about privacy. Graph Search may make it possible uncover things you didn’t know where public. For example, it’d be very easy for a person to jump far back in your timeline with a search like “Photos taken of Andrew Sims in 1990.”

Facebook has not yet announced a wide opening date for this feature, but will presumably be slowly rolled out to users over the next few weeks.

You can watch a video demonstrating the feature below, and learn more on this section of their site (at the bottom you can apply for the beta).