Moscow. August 23. INTERFAX.RU – Employees of the St. Petersburg Federal Research Center of the Russian Academy of Sciences trained artificial intelligence to identify groups of malicious bots in social networks, the press service of the city governor's administration said on Monday.
“The method is based on the analysis of public data about bots, regardless of the language in which they write posts and comments,” the message says.
Scientists study the activity of bots in social networks, their interaction with each other and with users. The data obtained makes it possible with a high degree of probability to understand which accounts belong to people and which are bots.
“To train the neural network, we created special groups in social networks, into which we introduced bots, and of different quality – both simple ones and those that can disguise themselves well as real users. their disguise “, – said one of the developers, leading researcher at the Laboratory of Computer Security Problems of the St. Petersburg Institute for Informatics and Automation of the Russian Academy of Sciences Andrei Chechulin.
The method can be useful for companies that use social networks for commercial purposes to detect and counter information attacks. The development will help not only to identify bots, but also to assess their quality, to roughly calculate the cost of an attack.
“For example, we look at a restaurant's social network account, and there are a lot of negative comments. We can identify whether the bots left them or real people. If the bots, the restaurant will understand that it is being attacked. In addition, we can determine the quality and the capabilities of bots and understand how much money was invested in this attack. Based on this data, it will be easier for businesses to take measures to effectively respond to this attack, “Chechulin said.
The research results are published in the international journal JoWUA. The project was supported by a grant from the Russian Science Foundation.
