NEW YORK: Until now, training the Gmail filter by marking messages as spam (or “not spam”) mostly trained the overall machine learning algorithm, it seems, but didn’t have a lot of influence on your individual mailbox. However, its latest approach involves an artificial neural network, the same tech which the company deploys behind Google Now and Google Search. Google has a whole lot of details on how bulk senders of email can properly format their messages to make sure they’re sorted into the correct inboxes up on its support site, but interested senders can start analyzing their performance right away.
The mail you want, not the spam you don’t [Gmail Blog]. On top of this, Google is using the technology in the spam filter to better understand the different tastes of the users on Gmail. It’s now implementing machine learning to detect and block spam trying to pass off as wanted mail; Google says it’s smart enough to weed out phishing scams.
The AI will also learn about your individual preferences, too.
In other words, if user has subscribed to a newsletter, but never wants to read it, it sounds like it might hit the spam folder more frequently.
In a post on the Gmail blog, product manager Sri Harsha Somanchi wrote that less than 0.1% of the email in an average Gmail inbox is spam, and the false positive rate is even lower at just 0.05%.
Google also notes that the spam filter is now better than ever at rooting out email impersonation, which is where many phishing attacks come from. However, people who use Gmail still have to click the “not spam” button from time to time, which means that those users have to go through the entire spam folder in order to reach an email that is actually relevant but was somehow flagged as being spam.
The addition of Postmaster Tools could prove to be a very valuable tool for businesses having deliverability issues.
