Special artificial intelligence software, for better understanding of the issues for which users discuss their posts, developed by Facebook.
The Deep Text can understand text with “almost human accuracy” refers to related blogpost of Facebook, which complemented that the system is able to “catch” the contents of thousands of post per second, in over 20 languages.
The system is based on research papers of Ronan Kolompert and Jan LeKoun of Facebook AI Research.
“It is important for the Deep Text to understand as many languages as possible is … through the use of deep learning, we can reduce the dependence on knowledge depends on the language, and the system can learn from text with little or no treatment “refers about.
The Deep Text already tested some tools / Facebook’s services: If the Messenger, for example, is used by the group AML Conversation Understanding to be understood better when someone might want to go somewhere. Used ‘intention finding “something that helps to be understood that anyone not looking taxi when he writes” just stepped out of the taxi, “as opposed to” I need a taxi. “
Also, one could write a post in which he says that he wants to sell his old bike for $ 200. The DeepText would be able to identify that the post has to do with a sale, export the information substance as the object sold and price, and to urge the seller to use existing tools such transactions facebook. Even the Deep Text makes it easier to tackle spam, while finding the most “qualitative” comments in discussions involving large numbers of users (eg to posts celebrity).