Researchers at Keele University in the U.K. have created a tool that promises to help detect fake news with a reported 99% rate of accuracy — thereby potentially reducing the incidence of online misinformation.

To accomplish this, the researchers used different machine learning techniques to build their model, which scans news content to issue a judgment of whether a news source is legitimate or not.

According to the researchers, the approach used relies on an "ensemble voting" technique that marries the predictions of several different machine learning models to issue an overall score.

During trials, the technique successfully identified fake news 99% of the time — far exceeding the researchers' earlier predictions and expectations.

The team aims to further refine the technique so that they can produce a model capable of identifying fake news 100% of the time.

The Keele researchers explained, "In our constantly evolving digital communication landscape, the widespread dissemination of false information is a significant concern. It compromises the integrity of public discourse and has the potential to threaten both local and national security via influencing biased mindsets, views, and actions. The risk posed by misinformation, disinformation, or fake news to the credibility of online news platforms, particularly on social media, highlights the urgent need for innovative solutions. We aim to enhance the capabilities of our AI solution through further research to help nip this problem in the bud."

An article detailing the technique,An Ensemble Modelling of Feature Engineering and Predictions for Enhanced Fake News Detection,” appears in the journal Artificial Intelligence XLI.

To contact the author of this article, email mdonlon@globalspec.com