Dutch student's algorithm detects Twitter account hacks

A University of Twente student says she has developed an algorithm that can help Twitter and Facebook determine if an account on the social media platforms was hacked. Meike Nauta says her algorithm analyzes six data points that can determine a hack to a 99-percent certainty, reports Dutch media outlet Computable.

In the analysis, she determined that an algorithm combining the frequency of publishing Twitter messages, combined with the language of a Tweet, the time of day it was sent, the device used to send the message, the domains of any links in it, and if it is a re-Tweet can nearly guarantee if the account holder sent a message. The Business and IT undergraduate student used that knowledge to create an algorithm that analyzes user behavior.

As such, Nauta proposes that any social media platform can also utilize the algorithm.

“If you discover a hacked account within 24 hours, you can reduce damage by seventy percent, according to the literature. Thus, those tweets containing malicious links are quickly removed so that fewer people suffer,” Nauta told Computable. "If you use Twitter, you always get spam,” which she says Twitter normally removes quickly. “However, it is difficult to detect spam if it is sent from hacked accounts.”

Nauta researched Dutch Twitter accounts for users warning their followers that their Tweets were the results of a hack. She analyzed messages sent between 2013 and 2016, and found 18 thousand cases of people saying their accounts were compromised.

Nauta’s research was published and discussed at the Webist conference in Porto, Portugal last week.