Researchers at Germany's universities of Göttingen and Frankfurt and Slovenia's Joef Stefan Institute used machine learning to develop a method for identifying fake news even when such reports are repeatedly adapted.
The approach creates classification models that identify suspicious messages based on content and certain linguistic characteristics, such as comprehensibility and mood.
With the new model, which is similar in principle to spam filters, fraudsters cannot evade detection by adapting their messages to avoid certain words. This means that even if "suspicious" words are removed from the text, its linguistic features can still identify it as fake news.
Said Michael Sierling of Goethe University Frankfurt, "This puts scammers into a dilemma. They can only avoid detection if they change the mood of the text so that it is negative, for instance. But then they would miss their target of inducing investors to buy certain stocks."
From University of Gottingen (Germany)
View Full Article
Abstracts Copyright © 2021 SmithBucklin, Washington, DC, USA