Facebook is getting close to the human understandings with their new artificial Intelligence DeepText.
Facebook knows the user status what he is posting on the timeline or in the messenger. DeepText already tested in the messenger, if we consider an example and know the difference between the contexts “I just got out of the Taxi” and “I need a ride” messages. Then DeepText analyses the context and it asks if you want an Uber. If a user is trying to sell a product and if it detects, then it automatically formats the post with the price and product details.
Facebook can determine what kinds of comments on celebrity posts are interesting and not just “Oh My Gosh I Love You”, it can surface one you want to read the text. These comments to the certain posts are relevant and high-quality comments that should be come in the threads attached to the celebrity posts and or the Facebook live streams.
Each minute more than 400,000 new stories and more than 125,000 comments are generated from the users. Unlike other systems Facebook’s DeepText analyses these thousands of posts and messages across the different languages with the most accuracy rate as the humans.
DeepText was going to better at understanding text thanks to how many words that people were publishing on the Facebook, Messenger, WhatsApp and the Instagram as well as the other which Facebook uses for the cross-reference with the selected text.
If we consider the image as an example in the Facebook, DeepText uses the Facebook’s Image recognition technology to identify that if someone captions an image of a baby with the caption as “Day 20” which is most likely means the baby is 20 days old.