New way to expose untrustworthy sites devised by scientists to combat fake news
17 April 2020, 00:04
Researchers say looking at third party platforms, such as video hosts, can be an indicator of whether a site can be trusted.
The type of video streaming or advertising platform used by a website could be used as a marker to expose those producing fake news, researchers claim.
While website owners host the pages, the videos and advertising end is usually served by a third party.
Scientists from the University of Warwick said the new approach may help tech giants flag untrustworthy articles more rapidly and prevent their misleading content going viral.
It comes amid heightened concern about fake news, as people seek information about the ongoing coronavirus pandemic.
Some 450 top news websites, as ranked by web traffic firm alexa.com, were analysed against 50 fake news websites and 50 clickbait sites identified by the Harvard University Library.
They spotted 115 significant third parties that were only used by trustworthy sites and seven that were only used by untrustworthy platforms.
This could therefore be used as a marker to recognise untrustworthy websites with an accuracy of 94%, scientists say.
“The US presidential elections in 2016 highlighted the significant harm that fake news can do, potentially impacting election outcomes and undermining democratic institutions,” said Ram Gopal, professor of information systems management at Warwick Business School.
“These concerns have multiplied during the coronavirus pandemic and fake news has resulted in an untold number of deaths from misleading and harmful information.
“It is vital that we use all the tools at our disposal to combat the spread of fake news and the huge damage it does.”
The group believe their technique could complement artificial intelligence, which can be fooled by quick changes to an article.
“Trying to identify fake news articles is a cat and mouse game, because the content can be quickly changed to defeat the algorithms searching for them,” Professor Gopal continued.
“To detect fake news effectively we need strong markers that are difficult to hide or fudge.
“A website’s choice of third-party partners exposes the essence of what the website does and how it achieves that. A tiger cannot hide its stripes.”
Their findings were published in the ACM Transactions on Management Information Systems journal.