Graph Search: a recommendation engine only Facebook could power
Facebook’s new Graph Search is unlike any other search engine. It opens the vault of likes and interests stored in Facebook and combines it with meaningful social identity.
In other words, it’s something only Facebook could do.
When people sign up for the social network, they list their age, gender, relationship status, hometown, current city, schools, workplace and much more. Every subsequent action people take on Facebook is connected to that identity, so when someone Likes a movie or checks into a restaurant, Facebook starts to get a picture of the type of person who is a fan of a particular movie or goes to a certain restaurant.
Now with Graph Search, that information will be accessible to the average user, and it will be able to be uncovered in reverse. It’s one thing to be able to see that a movie is liked by women between the ages of 16 and 36 who also like American Idol. It’s another to enable users to search for “movies liked by women between 16 and 36 and who like American Idol.” That’s what Graph Search does.
This is a query Google and other search engines can’t answer because they typically crawl the web for keywords. Doing just that, Google returns results for Kelly Clarkson’s Wikipedia page, articles about the current season of American Idol and a story about the Miss America pageant choosing the final 16 contestants. Even Google’s Knowledge Graph, which seeks to answer users’ questions directly instead of leading them to another site, is limited because it lacks the personal identity details that Facebook has.
There are a number of new and interesting use cases for Graph Search. It can be a people finder, a market research tool, a navigation system for photos on the site, and more. One of the most compelling ways to use Graph Search is as a recommendation engine. Users can combine a number of properties to find music, movies, TV shows, restaurants, tourist attractions, games and other things that they are likely to be interested in.
Friend recommendations can be quite meaningful, but in other situations, users might value the opinion of other groups of people.
For now, Graph Search only indexes people, places, photos and interests, but as the company integrates Open Graph and status updates, it will become an even more powerful recommendation engine.