Facebook engineers have a lot challenges ahead of them as they work on the social network’s new search tool—the Graph Search. One of their many challenges is the over-abundance of data to sift through.
For instance, a user is searching for a Korean Restaurant in San Francisco, California liked by people from South Korea. The query, which is expected to generate hundreds of results, would only return two measly establishments.
Still in beta, the Facebook Graph Search simply does not have the processing power to sift through the million of connections among Koreans on the site to conduct the search. That is according to Facebook engineers during a small media briefing last Thursday at their headquarters in Menlo Park, California.
In addition, software engineer Michael Curtis noted there’s still a lot of work that they need to do. Computationally speaking, a long query is very difficult, as it is virtually intractable with the limited amount of time that the search engine has. Thus, they end up cutting out possibly good results.
Curtis helped design the social network’s Unicorn search engine that provides Graph Search’s infrastructure.
Various Approaches in Search of a Solution
To solve these big data problems associated with Graph Search, Facebook is taking various approaches. One of them includes a concept in computer databases known as “query optimization”. This is meant to improve the speed and efficiency of certain types of searches.
In the case of the Korean restaurants, the technique could be applied to start first with the restaurants that are liked instead of starting with Korea. And then it will be filtered down the liked by people, Facebook engineers said.
Facebook is also addressing the challenges at a hardware level by adding flash memories, as well as other new features to the servers they use as data centers. That is in order to accommodate the increase in search traffic caused by Graph Search. As stated by Soren Lassen, the lead of Graph Search’s search infrastructure team:
We need to do extra work in data centers, buying new hardware platforms, [with] new types of servers being put up to support the computational needs of Unicorn.
Facebook started rolling out the Graph Search to a limited number of users in the US last January. The search tool is designed to let people comb through the site’s 1 trillion connections to search for people, places, photos, and interests using English phrases.
Graph Search would eventually incorporate other metrics such as user comments and status updates to compile and rank results. But that’s still in the horizon.