Facebook’s novel search method improvement
A few years ago, Facebook made a novel experiment related to network. A new search tool, Graph Search was online. This enhanced version of search tool uses information gathered from friends, friends of friends to help users make better decision. This function emphasizes on the friends’ preferences on Facebook. Users can type a short phrase like “books my friends like”, Facebook will display specific books on which users’ friends click the Like button. This pattern also works for restaurants or music.
Another function is to find your old friends. For example, when users want to find one friend named Mike, although there are too many Mike on Facebook, it can be done by simply entering “people named Mike who are friends of my friends.” As long as user’s friend Mike has some mutual friends with him.
For the first function I mentioned in the first paragraph, Facebook generates a network which treats users’ friends as nodes and checks connectivity of this network. It reflects every nodes’ preferences when the user conduct one search. For the second example of the function, it illustrates Triadic Closure, if Mike and the user both in one social network(Facebook) and have a friend in common, chances are very high that they will become friends or in that case, they already knew each other. Beyond that, the strong triadic closure property is also important when users do a graph search. Facebook can find friends of friends’ preferences based on the fact that if node A has two good friends B and C, they are likely to form an edge. So when B make a graph research, Facebook can use C’s data to make a recommendation and vice versa.
In conclusion, Facebook’s Graph Search is a novel experiment of network search technology. Every node in a giant component can share their hobbies and preferences based on this new search method.