In the past few months since Facebook launched its Graph Search function in beta in January, just a handful of my friends were permitted to see the the capabilities of the tool, which exposes people, places and things and the connections among them that are otherwise hidden away in individual Facebook profiles. But now lots more people are about to get a better of sense of what the heck a graph is and what it can do. On Monday the social networking company said it was rolling out the tool to all users who use Facebook in United States English.
As I have written before, there are limits to the feature. But I must admit I’m excited, partly because now lots more people will be able to see for themselves how the graph model for data can efficiently surface connections that otherwise would take a long while to pop up if it were all in a relational database.
Plus, others will be able to consume data in Facebook that previously was buried. For example, a few months ago I logged onto Facebook and typed into the Graph Search box “friends in San Francisco,” which pointed to a bunch of Facebook friends I’d forgotten about who now live in the same city as I do. I’ve since caught up with a few of them. And it’s not like finding this out took a long time: the results came up right away. Facebook engineers have thought up hacks to take in user input and spit back data almost immediately, which is one reason why why Graph Search works so fast.
Also, anyone who Graph Search can get a taste of natural-language processing, an increasingly popular field that aims to understand information the way regular people put it in. Graph Search puts it to use by turning user input into search strings the Unicorn search engine can easily grasp. It cleans up grammar, accepts slang and takes educated guesses about the sorts of entities that users have in mind when they search.
Machine learning is important, too. it improves results over time with the help of user feedback.
Facebook admittedly still has work cut out for itself with Graph Search — working it into users’ mobile experiences, incorporating posts, comments and other data into searches, and making it available in languages other than English. But at least I don’t have to use my own words to explain it to friends. They can try it out for themselves.