Starting from 2004, Facebook continues to grow far and wide in terms of time people spent on the platform and number of active users at any time of the day. The internet media giant has 2.45 billion monthly active users.
The major issue that the company had been facing for years is the increasing number of fake accounts. Finally, Facebook has revealed it’s a key technology that has enabled it to handle the massive chunk of fake accounts used for everything from the spread of false information to spam as campaigns.
On this Wednesday, the social media giant disclosed details of how its AI researchers have developed an artificial intelligence system and prepared it to be error-free to detect accounts that infringe its policies automatically.
Need For Eliminating Fake Accounts
Regulating its giant social network has become an increasingly experiential issue for the company as it faces the increasing threat of regulation nationwide. The lawmakers and public have been greatly disappointed by the role the Internet media giant has played, ranging from Myanmar’s genocide against the Rohingya Muslims to Russian interference in the 2016 US Presidential election.
As the 2020 US presidential election approaches, it becomes exceptionally pivotal for Facebook to acknowledge social media influence on voting behavior, new narratives, and other key election matters. Trying to scam users and promoting spam is just one side of the problem of the fake account on Facebook. Users and government officials have also become warned about phishing, hate speech, financial fraud, and bullying effectuated on the platform.
What Was The Conventional Way Of Dealing With This Menace?
Before the artificial intelligence came into picture, Facebook largely depended on users to report the offending accounts to workers who were deployed only to review the accounts. However, depending on human reviewers has generated other issues too. The contract workers at Facebook were used to review suspicious behavior and content, but these men were often paid low, and due to continuous exposure to malicious posts, videos and images suffered from mental health problems.
Further, the number of problematic accounts that Facebook has to deal with is massive. In the third quarter of 2019, Facebook has blocked some 1.7 billion suspicious accounts. At any moment, Facebook approximates that 5% of its active accounts are fraudulent.
The New AI Technology and It’s Working
The company has a very new artificial intelligence-powered technique called Deep Entity Classification or abbreviated as DEC.
This machine learning method was designed to deal with the limitations of conventional methods of identifying fake accounts, which usually focus on the user’s profile.
The new technique addresses existing issues by considering other features as well. It detects a suspect account by looking to all the pages, friends, and groups with which it interacts. It then explores all of these entities. The pages can be inspected by the number of admins it has, whereas friends can be evaluated by the group they are part of, number of friends, and ages. Groups can be evaluated by the admin and number of members it contains.
Following the inspection of properties and behaviors of these entities, the method gets a strong message about whether an account is fake or not.
However, if the investigation turns out to be wrong, the accused get a chance to justify as their account enter an appeal process before they are banned.
The company has recognized that is DEC technique needs to be continuously upgraded to ensure that it remains productive against the spammers’ always changing tactics. It is these types of techniques that will keep Facebook hard at work, trying to stay one step ahead.
An engineer and a tech enthusiast, driven by curiosity.