Facebook's Artificial Intelligence Understands You

— -- If there are days when it feels like no one really understands you, take solace in knowing Facebook's artificially intelligent DeepText engine does.

In a blog post sharing new details about the engine, Facebook's engineering team said DeepText can understand "with near-human accuracy the textual content of several thousands posts per second, spanning more than 20 languages."

How It Works

Traditional Natural Language Processing converts words into numbers, which can then be read by a computer algorithm. Its pitfall, however, is not preserving the semantic relationship between words, such as "brother" and "bro," which would both be assigned different numbers.

DeepText is able to understand more natural language because it uses "word embeddings," a concept that keeps the relationship between words that mean the same thing but might have some derivatives.

Why It Matters

DeepText is designed to make a Facebook user's experience even easier by connecting them to the tools they may need at that moment and the posts that are most relevant to their interests.

"For example, someone could write a post that says, 'I would like to sell my old bike for $200, anyone interested?'" the blog post said. "DeepText would be able to detect that the post is about selling something, extract the meaningful information such as the object being sold and its price, and prompt the seller to use existing tools that make these transactions easier through Facebook."

It's already being tested in some parts of Facebook, including Messenger, where DeepText can determine intent. One example is if the user writes they need a ride, it could help them order a car service. DeepText is smart enough to recognize intent, so it wouldn't offer the same help to someone who may write to a friend: "I just came out of the taxi."

DeepText can also filter spam or abusive comments, making it easier for public figures and their fans to engage on Facebook.

"Finding the most relevant comments in multiple languages while maintaining comment quality is currently a challenge," the post said. "One additional challenge that DeepText may be able to address is surfacing the most relevant or high-quality comments."