Now publications that violate the rules of the social network (spam, incitement to hatred, promotion of violence
Now about 15 thousand people work on Facebook. moderators around the world, but users and the media criticized them for the fact that the site administration does not support them enough and does not hire more staff. Their job is to sort through the complaint posts and make decisions about whether they are breaking company rules or not.
Previously, moderators reviewed posts based onPosts - Facebook has now changed its approach to prioritize posts that gain more reach. The artificial intelligence (AI) model will use a set of criteria - virality, the subject of the post and the likelihood that it breaks the rules.
Violating Facebook groups face probation and forced moderation
Facebook notes that the new approach will helpunderstand the most common posts that may violate community rules. The social network also notes that it flags posts with “potential harm” so that they can be reviewed first. This will help you promptly remove content related to terrorism, child exploitation, or self-harm. Spam will be considered only later - they note that such messages are annoying, but cannot injure users.
Facebook also shared details of howMachine learning filters work - the model holistically evaluates user content on a social network. This means that algorithms jointly evaluate various elements in any post, trying to figure out what the picture and its caption mean as a whole.
Read also
Hidden gene responsible for pandemic found in genome of coronavirus
The Doomsday glacier turned out to be more dangerous than scientists thought. We tell the main thing
It turned out that the universe is heating up. Temperature has increased 10 times in 10 billion years