Communications of the ACM
Wpi Research Detects When Online Reviews and News Are a Paid-For Pack of Lies
Worcester Polytechnic Institute researchers have developed algorithms designed to combat online workers' posting of phony reviews, circulation of malicious tweets, and spreading of fake news.
Credit: iStock
Researchers at Worcester Polytechnic Institute (WPI) have developed algorithms designed to combat crowdturfing, a phenomenon in which masses of online workers are paid to post phony reviews, circulate malicious tweets, and spread fake news.
The researchers, led by WPI professor Kyumin Lee, say the algorithms are highly accurate in detecting fake "likes" and followers across various platforms.
Lee focuses on crowdsourcing sites such as Amazon's Mechanical Turk, and although he says most of its tasks are legitimate, the site is sometimes used to recruit people to work on crowdturfing campaigns.
The researchers used machine learning and predictive modeling to create algorithms that sift through the posted tasks seeking patterns associated with illegitimate tasks.
The algorithm can identify the organizations posting the tasks, the sites the crowdturfers are told to target, and the individual workers who are signing up to complete the tasks. The algorithms can detect fake likes with 90% accuracy and fake followers with 99% accuracy.
From Worcester Polytechnic Institute
View Full Article
Abstracts Copyright © 2017 Information Inc., Bethesda, Maryland, USA
No entries found