Facebook Turns to Deep Learning to Grow Search Possibilities

Facebook is looking to expand its Graph Search algorithms to account for the use of slang and expressions, and translate them into searchable key words. The company adjusted its search algorithms in January, and the changes mostly worked. The new algorithms signal the beginning of new ways to search user information using natural and nuanced language, and provide more personalized ads. Other tech companies are looking into similar technologies.

“Like Google and Apple and other tech giants, Facebook is exploring a new field called ‘deep learning,’ which will allow its machines to better understand all sorts of nuanced language and behavior that we humans take for granted,” reports Wired. “In short, deep learning teaches machines to behave more like the human brain.”

When Graph Search launched, it was only able to search content between users and other groups, but now all activity on Facebook is accessible to Graph Search. Now search algorithms can be smarter, and Facebook will have more ability to use more natural language.

“Humans differ in the way they use language because of differences in their cultural upbringing. We still need to teach machines these nuances,” explains Oleg Rogynskyy, CEO of text analytics company Semantria. “Right now, there’s no way a machine can understand these things that precisely because it lacks the cultural context. That’s going to be the hardest thing to crack in the next 10 to 15 years.”

Deep learning is based on neural networks, which are multi-layered software systems modeled after the human brain. These artificial networks can obtain and respond to information, and can form understandings how objects appear, sound like, or what words mean without human labeling.

Companies such as Microsoft, Google and the Chinese search company Baidu have all used deep learning in their image and voice search. The next obstacle is to decode an individual’s written language and emotions.

But the smartest search algorithms are limited in their capability to discern accurate information as many currently take “bag of words” approaches, where search models do not consider word order and only see a mixed collection of words. Other similar algorithms consider strings of different word lengths, but are slightly more successful. These approaches work best when focusing on users as a whole, but companies want to target individuals with personalized messages and ads.

“That’s why Facebook and others are turning to deep learning,” suggests Wired. “They want technology that lets them better understand how individual users feel about and interact with, well, everything. They can use that information to improve user experience, build brand loyalty, and, ultimately, sell people stuff — all in a more finely tuned way than what’s currently possible.”