The biggest social media companies in the world have a lot to deal with on a daily basis. Abusive users, accounts promoting violence against certain groups, fake news manipulation and of course, fake accounts. Today, it has been revealed that a Facebook AI has been used to remove more than six billion fake accounts in the last year.
According to the company, this has been a monumentally important tool they’ve been employing in the past two years and their number one tool in fighting fake accounts. It uses a technology called “Deep Entity Classification” to analyse patterns within active Facebook accounts and individual profile behaviours in the entire community. The machine learning methodology has identified certain behaviours and indicators within the data to ascertain whether active accounts are fake, and deletes them accordingly.
As machine learning algorithms have been designed to work, it can learn over time without any additional human input. It simply needs more data to be more accurate in its predictions. It uses tens of thousands of data metrics to make its decision, including obvious Facebook data like the number of friend requests an account has sent and the groups or pages it has joined or followed.
In an interview with ZDNet, Facebook data science manager Bochra Gharbaoui said that the tool they’ve built has reduced the number of spam accounts by around 27 percent. That still leaves Facebook with a whopping 5 percent of its accounts being spam or fake accounts.
Of course, Facebook is fighting fire with fire here. Just the sheer number of fake accounts on the platform is baffling. Spammers use similar machine learning methods to build and run its fake accounts for various reasons, some more nefarious than others. As you can imagine, with an active user base of over 2 billion people, Facebook’s fake account problem isn’t going anywhere anytime soon.