Twitter Inc (NYSE:TWTR) Social bots are polluting the platform with spam, but the team lead by Emilio Ferrara at Indiana University in Bloomington now have a trick to track the social bots and segregate them from ordinary human users, according to a report from Technology Review. The team took help from the earlier database created by the Texas team, picking a set of social bots identified in 2011.
Social bots more advance now
In 2011, the team from Texas A&M University found a fix to catch such non-human accounts. The team decided to set up “honeypot” accounts, which used to post senseless content that would not interest a human user.
The team setup 60 honeypots and caught around 36,000 potential social bot accounts. Lots of people were taken by surprise after seeing the number of fake accounts, which were active. The non-human users were generally unsophisticated and simply retweeted more or less any content they came across.
Investment strategies used by hedge funds have evolved over the years, although the biggest changes have come in the use of computers to develop portfolios. Rosetta Analytics is a woman-founded and woman-led CTA that's pioneering the use of artificial intelligence and deep reinforcement learning to build and manage alternative investment strategies for institutional and private Read More
In a present time, social bots are comparatively more advanced. Now these accounts track the popular and influential people on social networks like Twitter, and send them messages. Also, these bots are smart enough to recognize the keywords and search for the content accordingly, and some were advanced enough to answer using the natural language algorithms. It has become even more difficult to hunt down such social bots on social platforms like Twitter.
Difference in the Twitter is used by bots
Ferrara’s team chose 15,000 social bot accounts from the 2011 list, and collected their 200 most recent tweets, and 100 most recent tweets mentioning them. This provided them with some 2.6 million tweets. Same way they collected over 3 million tweets from 16000 human users.
The researchers developed an algorithm titled as Bot or Not, a program that digs deeper into the differences between the properties of human and social bots. Also, the algorithm has taken into account 1,000 features associated with these accounts like the number of tweets and retweets each user posted, the number of replies, mentions and retweets each received, the username length, and even the date when the account was created.
Upon analyzing the details, it was found that there are some significant differences between the human accounts and bot accounts. Bots are more frequents in tweeting compared to the humans and also they have younger accounts, whereas human receive more replies, mentions and retweets.
“Bot or Not?achieves very promising detection accuracy,” say Ferrara and pals.