Facebook communicates for the first time statistics on their content blocking

Facebook communicates for the first time statistics on their content blocking
geralt / Pixabay

Facebook published today, for the first time, statistics on the number of publications blocked on the social network between October 2017 and March 2018. The figures cover six areas: graphic violence, adult nudity and sexual activity, terrorist propaganda, hate speech, spam and false accounts.

The figures show:

  • The amount of content that people have seen and that violates the Facebook standards
  • The amount of content removed
  • The amount of content detected proactive way-before Facebook users report it.

Spam

Most of the steps taken to suppress bad content are spam and fake accounts used to distribute them. Like what:

837 million spam was deleted in the first quarter of 2018, of which almost 100% were detected before anyone reported it.

The key to the fight against spam is to remove the fake accounts that propagate it. In the first quarter, about 583 million of false accounts were deactivated, most minutes after registration. This is in addition to the millions of attempts to fake accounts prevented daily from registering on Facebook. On the whole, Facebook estimates that about 3 to 4% of the Facebook accounts active on the site during this period were still false.

Other types of non-compliant content

21 million pieces of adult nudity and sexual activity were removed at T1 2018-of which 96% were found and reported by an automatic algorithm before these publications were reported. On the whole, Facebook estimates that out of 10 000 content elements viewed on the social network, 7 to 9 views were content that violated standards for adult nudity and pornography.

With respect to graphic violence, 3.5 million of elements of violent content were suppressed or censored at T1 2018-of which 86% were detected by an automatic algorithm before these publications were reported.

In terms of hate speech, the technology still does not work properly, admittedly by Facebook, and therefore must be verified manually. 2.5 million hate speech elements were suppressed in the first quarter of 2018-of which 38% were reported by automatic algorithms.

 

Source: Facebook Newsroom

You may also like...

Leave a Reply

Your email address will not be published. Required fields are marked *