Your latest ‘like’, ‘retweet’ or ‘follow’ may not have been from a human, a new study has found.
Researchers have discovered that up to 15 percent of Twitter accounts are bots and with 319 million active users on the site, it is estimate that 48 million of them are computer programs.
While some bots can wreak havoc on Twitter, sources have revealed that ‘many bot accounts are extremely beneficial’, as they ‘alert people of natural disasters or from customer service points of view’.
Researchers have discovered that up to 15 percent of Twitter accounts are bots and with 319 million active users on the site, it is estimate that 48 million of them are computer programs. Human users are represented in blue and bots are shown in red
The recent investigation into Twitter was conducted by the University of Southern California, which used 1,150 features from six classes to track down accounts run by bots.
This included data like tweet content and sentiment, network patterns and activity time series.
‘We benchmark the classification framework by using a publicly available dataset of Twitter bots,’ the researchers shared in the study published in arXiv.
‘This training data is enriched by a manually annotated collection of active Twitter users that include both humans and bots of varying sophistication.’
BOT DETECTION FRAMEWORK
The team used over 1,150 features in six different classes to identify the bot accounts.
The classes are:
User-based features: Number of friends and followers, the number of tweets produced by the users, profile description and settings.
Friends features: Retweeting, mentioning, being retweeted, and being mentioned.
Network features: Retweet, mention, and hashtag co-occurrence networks.
Temporal features: Average rates of tweet production over various time periods and distributions of time intervals between events.
Content and language features: Looked at the nature of social media conversations.
For example, deceiving messages generally exhibit informal language and short sentences.
Sentiment features: Used several sentiment extraction techniques to generate various sentiment features, including arousal, valence and dominance scores, happiness score, polarization and strength and emoticon score.
Researchers were able to identify nearly 14 million bot accounts on Twitter using their system and optimal threshold scores that separate human and bot accounts. During the analysis, the team also discovered the bots exhibited different types of behavior
‘Our models yield high accuracy and agreement with each other and can detect bots of different nature.’
‘Our estimates suggest that between 9% and 15% of active Twitter accounts are bots.’
However, sources told DailyMail.com that ‘while bots generally have negative connotations, many bot accounts are very beneficial, like those that automatically alert people of natural disasters (flash floods, earthquakes, tsunamis, air quality, etc) or from customer service points of view’.
The researchers also explained how these bots can operate for the great goods as well.
‘Many social bots perform useful functions, such as dissemination of news and publications and coordination of volunteer activities,’ they explained.
However, the team also noted that some of the computer programs have the ability to ‘emulate human behavior to manufacture fake grassroots political support, promote terrorist propaganda and recruitment, manipulate the stock market and disseminate rumors and conspiracy theories’.
THE STAR WARS TWITTER BOTS
University College London found more than 350,000 Twitter bots working as a team.
And they have been named ‘Star Wars bots’ because they only posted random lines from the Star Wars novels.
A seperate study discovered a botnet that gathers the force of more than 350,000 bots. Dubbed Star Wars bots, these accounts were discovered in a small sample of the site and were found to only tweet random lines from Star Wars novels
Many of the quotes began or ended with an incomplete word.
There are also hashtags inserted in random parts of the post.
The bots posted no more than 11 tweets and max 10 followers and 31 friends. Experts believe the bots were also created using a Windows smartphone.
However, all of these accounts went quite on July 14, 2013, which suggests the bots were orchestrated and controlled by a botmaster.
When the team plotted the locations of these accounts on a world map, they discovered 3,244 of them were within two rectangles in North American, Europe and parts of North Africa.
Researchers were able to identify nearly 14 million bot accounts on Twitter using their system and optimal threshold scores that separate human and bot accounts.
During the analysis, the team also discovered the bots exhibited different types of behavior.
‘Simple bots tend to retweet each other, while they frequently mention sophisticated bots,’ reads the study.
‘More sophisticated bots retweet, but do not mention humans.’
‘They might be unable to engage in meaningful exchanges with humans.
‘While humans also retweet bots, as they may post interesting content, they have no interest in mentioning bots directly.’