How Facebook is able to take your Likes and turn it into contextual data
Facebook is now up to 900 million users, which was uncovered in their recent Initial Public Offering. Having that many users, close to a billion, is astronomic for any one entity which then has to take that user data and turn it into something - that something being useful data which is forming the social web which many want to be a part of and which Facebook can benefit financially from. So, how does Facebook do it all?
Facebook is able to obtain and make use of data via the Facebook Platform, which is the developers platform where you can integrate directly into Facebook via their tools. Facebook is so big, Wired said that they have more than 9 million apps and websites which are integrated directly into Facebook, which sends data back to the social network to analyze. The data is then managed by Facebook using several tools of their own.
The two main tools that Facebook uses to consume and make sense of all the data is their Social Graph and their Open Graph, which uses “objects” such as a photo, and determines the relationships between objects. Software Engineer Ripu Jain summarized how Facebook uses these two tools gracefully in his Google+ post:
Social Graph: data model representing the connections between its million of users.
Open Graph: means of grabbing all sorts of information people generate on third-party services and feeding it back into Facebook data centers.
In Open Graph, there is an Object Store, which stores things like users and events and groups and photos, and then theres an Edge Store that stores the relationship between objects.
One problem is understanding globally how people interact with this content. The secondary problem is trying to understand on a per user basis what is most interesting to them. If you prefer music, FB shows you more music. If you prefer games, FB shows you more games.
Then FB merges those two sets of scores together, to influence what Newsfeed shows and what Timeline shows and what some other systems show.
As 900 million people use FB, clicking on objects here and there, the behavior is recorded in a software platform called Scribe, a technology specifically designed to log large amount of data in realtime.
Then a second (unnamed) platform taps into Scribe and does a kind of on-the-fly analysis of this data, determining what’s the most popular and what’s the least.
The tallies are then shuttled to Facebook’s Newsfeed and Timeline platforms, and there — in tandem with a similar analysis of your personal behavior — they’re used to determine what Open Graph data you see and what you don’t.