Facebook's DEC AI Can Remove Billions of Fake Accounts: Report
Sat, April 10, 2021

Facebook's DEC AI Can Remove Billions of Fake Accounts: Report

Among all the social media platforms, Facebook is probably the most popular and most used. Its services have attracted billions of loyal users across the world. As of the third quarter of 2018, Facebook had over 2.375 billion daily active users – an 11.2% increase year on year / Photo by: Patrick De Grijs via 123RF

 

Among all the social media platforms, Facebook is probably the most popular and most used. Its services have attracted billions of loyal users across the world. As of the third quarter of 2018, Facebook had over 2.375 billion daily active users – an 11.2% increase year on year. Every day, the platform adds 500,000 new users, which is equivalent to six new profiles every second. Worldwide, 26.3% of the online population use Facebook.

Web Hosting Rating, an online site that covers unbiased, honest, and expert web hosting reviews, states that more women than men use Facebook with 74% of women compared to 62% of men. It was reported that Facebook is accessed on average of eight times per day, followed by Instagram (6), Twitter (5), and Facebook Messenger (3). Also, around 25% of Facebook users are between 25 to 24 years old. This is the age group using Facebook the most.

However, with the billions of people using Facebook every day, not everyone is real. During late 2017 and early 2018, Facebook detected and suspended some 1.3 billion fake accounts. However, an estimated 3% to 4% of fake accounts remain undetected. This is equivalent to approximately 66 million to 88 million profiles. At the same time, it is estimated that 9% to 15% of Twitter’s 336 million accounts are fake. 

According to Phys, an internet science news portal, fake accounts are not only present on Facebook and Twitter. In December 2017, German intelligence officials warned that Chinese agents are using fake LinkedIn profiles to target more than 10,000 German government employees. The Israeli military reported in mid-august 2018 that Hamas was using fake profiles on Facebook, Instagram, and WhatsApp to entrap Israeli soldiers into downloading malicious software. 

Why Fake Profiles Are Dangerous

In the most recent Community Standards Enforcement Report, Facebook showed that it removed more than 3.2 billion fake accounts between April and September 2019.  Many people are questioning how they were even created. 

Previous reports showed that most fake social media accounts are “bots.” They were created by automated programs to post certain kinds of information, which are mostly political. Sophisticated and more advanced programs can create millions of accounts. This is a violation of Facebook’s terms of service and is an effort to manipulate social conversations. 

Fake accounts are being used to amplify the popularity or dislike of a person or movement. Researchers have discovered that bots played a huge role in spreading misinformation on social media ahead of the 2016 US election. They aim to sow distrust and social division across the world. In some cases, they have even initiated violence against individuals or groups. 

Kathleen Carley, a computer scientist from Carnegie Mellon University, stated that aside from manipulating the conversation, they can also build and bridge groups. "They can make people in one group believe they think the same thing as people in another group, and in doing so they build echo chambers,” she said. 

In the most recent Community Standards Enforcement Report, Facebook showed that it removed more than 3.2 billion fake accounts between April and September 2019.  Many people are questioning how they were even created / Photo by: rawpixel via 123RF

 

How Facebook Detects Fake Accounts

Facebook has been criticized by many due to its billions of fake accounts. Aaron Greenspan, a Facebook critic, even claimed that half of the social media giant’s users could be fake. While Facebook debunked this claim, it acknowledges the fact that its fake account issue is not a good thing. “When it comes to abusive fake accounts, our intent is simple: find and remove as many as we can while removing as few authentic accounts as possible,” Alex Schultz, vice president of Analytics at Facebook, said. 

To address this issue, Facebook uses a machine learning framework called deep entity classification (DEC) to detect and remove fake accounts on its platform. In over two years, it has been responsible for a 20% reduction in abusive accounts on the platform, which concretely amounts to “hundreds of millions” of accounts. DEC has helped Facebook solve the problem of limited labels by employing a multistage, multitask-learning paradigm that leverages a large number of medium-precision, automated labels and a small number of high-precision, human-labeled samples. 

Facebook has been criticized by many due to its billions of fake accounts. Aaron Greenspan, a Facebook critic, even claimed that half of the social media giant’s users could be fake / Photo by: Ken Wolter via 123RF

 

“Over the past few years that DEC has been in production, we’ve seen a steep reduction in the number of [abusive] accounts on the platform. Even though attacker volumes increase, DEC catches them at pretty much the same volume,” Facebook software engineer Sara Khodeir said. 

According to VentureBeat, an American technology website that publishes news, analysis, long-form features, interviews, and videos, DEC was created to address problems Facebook encountered in its traditional approaches to automated fake account detection. It first considers the account’s direct features by entity types, such as age and gender (user entities), fan count and category (page), member count (group), operating system (device), and country and reputation (IP address). After that, DEC will consider looking into other entities the account interacts with, like pages, admins, group members, users sharing a device, groups shared to, and registered accounts.

Facebook’s DEC has played a huge role in detecting fake accounts. But aside from that, it has also contributed to making Facebook a platform with authentic users, removing those that want to manipulate public conversations.