June 2, 2016 report
Facebook introduces DeepText, text understanding engine
In its blog post, Facebook claims that DeepText is now able to read a blog post and understand what is in it, with "near-human accuracy." This is an exaggeration, of course, as the authors of the post then continue to outline the mechanics behind their project, the difficulties they have encountered, and challenges that yet remain.
The science behind the project, as its name implies, involves making use of several deep neural network architectures, each of which is based on the concept of multiple hidden layers that are able to "learn" by adding new bits of information as nodes to a network, similar in many respects to natural neural networks. Facebook has made clear that its goal is to one day be able to read a blog post or message , as it is being made, in order to provide a better experience for users. If a person writing to someone else via Messenger makes a comment about wanting to buy a certain pair of shoes, for example, than Facebook will be able to interrupt to offer a means for getting that done. Similarly, if someone is in the middle of writing a blog post about needing a ride to an event, they may find an ad for Uber popping up.
Everyone knows that Facebook staff can read blog posts at will, regardless of privacy controls put in by users, and that some can be censored if deemed inappropriate—this means, of course, that no blog posts are truly private—the same goes with messages sent in Messenger. But news that the company is planning to scan every message in real time might set off alarms regarding both privacy (which Facebook has heard before), or intrusion. With such software running in the background, users would most certainly come to feel its presence—a third party listening in during posting or while conversing, like a nosy neighbor on an old fashioned shared telephone line. It is not clear just yet if Facebook is preparing for such concerns, if users will tolerate such intrusions, or if there will be an option to turn them off.
© 2016 Tech Xplore