The social network Odnoklassniki has launched a technology based on neural networks to combat aggressive statements and inciting hatred in public comments. The neural network will evaluate the texts and images in the comments for toxic content, and then issue an assessment to the moderators for verification. The launch of the technology was another step in the development of a comfortable environment for communication in OK.
The new algorithm will create a respectful environment for communication within the social network. The neural network will automatically analyze the level of aggression in the comments, then the rating from the algorithm will be available to the moderators of the social network. The technology is already working on all platforms. In the desktop version of the social network, users who have received a temporary ban, when they try to leave a comment, will see a message with explanations and a link to details about the new algorithm. In the near future, this feature will appear in applications on iOS and Android.
All calls to incite hatred and xenophobia against any nation or other group of people, including using obscene language, as well as threats to other users, will be blocked. When analyzing images in comments, the algorithm will determine the presence of shocking content and acts of violence, which, after verification, will also be removed. The technology was launched as part of the global security building VK.
Users who violate the rules of the social network and express their opinions in a hostile manner towards others will receive a temporary ban on commenting on any content in OK. When you try to leave a comment, the social network will show a notification and a link to the rules of conduct. In addition, commenters can get a temporary “rest” from discussions if their comments are received a lot of complaints and the system has identified them as toxic and aggressive.
OK cares about user safety: in February 2022, the social network launched the Message Security Center, which contains the rules for safe and correct communication on the social network, as well as ways to protect against inappropriate content in personal correspondence.
#launched #neural #network #combat #aggressive #comments