VIP accounts on twitter with more than 10 million followers behave like bots. This is demonstrated by research conducted by the University of Cambridge
VIP accounts on twitter those with more than 10 million followers often behave like bots. This is demonstrated by recent research conducted by the University of Cambridge that has studied Twitter data to identify bots, study their behavior and understand what impact they have on social networking.
The researchers divided accounts into two categories, depending on the number of followers they had. They noted that profiles with more than 10 million followers tend to resign in a similar way to bots. The activities of the less followed profiles are more similar to human behaviors.
The results of this research will be presented at the International Conference IEEE / ACM International Conference on Advances in Social Networks Analysis and Mining (ASONAM) in Sydney, Australia.
Bot: malignant or benign?
Bots, like people, can be malicious. rude or respectful and polite. The term bot often associated with spam, offensive or partisan content. In reality, many large organizations use bots to convey their content. For example, the BCC or CNN who produce hundreds of content and news every day, rely on automatic sharing tools to publish in the most efficient way. These accounts, which users recognize as reliable sources of information, actually behave like bots.
A human Twitter user can behave like a spammer or a troll, as well as a bot can be benign and convey verified and useful information explains Zafar Galani, PhD student at Cambridge who participated in the research, interesting to understand how to identify automated accounts and study the effects their behavior has on the community.
It is estimated that out of the total number of Twitter accounts, there are lots of bots, between 40 and 60%. Some have tens of millions of followers, although most have less than a thousand. The numbers are similar to those of human profiles.
To identify the automatic profiles, the researchers used it online tool BotOrNot (later renamed BotOMeter), which turned out to be inaccurate. So the research saff decided to switch to a manual approach, visiting the profiles and determining if they were bots or not.
To classify profiles, students have taken into account several parameters: the date the account was created, the frequency of tweets, the content posted, the description on the profile, responses to other tweets, likes and interactions with friends. 3,535 accounts were analyzed: 1,525 were classified as bots and 2,010 as humans. They then developed an algorithm that proved to be accurate in 86% of cases and uses 21 different parameters.
What differentiates the behavior of bots from human behavior?
Bots tweet more than humans, tweet more often and frequently direct users to visit an external link. Exceptions are accounts with more than 10 million followers, where the differences fade and the behavior is similar, both in terms of tweets and retweets.
Probably because they cannot create original content and therefore prefer to rewrite the contents of others said Galani.
Furthermore bots, although they are becoming more and more sophisticated, fail to support structured conversations and put much less I like.
In terms of engagement, however, it is humans who record the most positive results. On average their tweets record 19 times more like and 10 times more retweets.
Many people think bots are evil, but not true. Just like people, they can be polite or rude. Now we want to study how their activity influences the community, what their social cost is, how do they modify relationships and conversations online? What we know for sure, that they will not disappear concludeGalani.