According to a study by employees at Cambridge University, Twitter accounts with a very large number of subscribers for behavior are similar to bots - even if they are driven by real people.

The researchers tried to determine how the bots behave on Twitter and how they can be identified. Employees of the university created an algorithm capable of evaluating accounts with an approximately 86 percent probability of detecting bots.

The algorithm was developed by analyzing various models of account behavior. He tracked indicators such as the number of retweets, the likelihood of a response to a message, and the frequency with which publications appear in links to third-party sites.

Employees of Cambridge University found that the behavior of accounts with more than 10 million subscribers are not very different from bots with the same number of subscribers. This, according to researchers, may be due to the fact that owners of large accounts like the news sites of the BBC and CNN automate the sending of at least some tweets for more effective dissemination of information.

As for bots and people with a small number of subscribers, the first in terms of the number of publications are significantly outstripped. "We think this is due to the fact that bots are not so good at creating original content on Twitter, so they rely much more on retweets and redirecting subscribers to external sites," said lead researcher Zafar Gilani,.

Among other things, the researchers found that tweets from real people get 19 times more likes and 10 times more retweets than the publication of bots.