Fake news is a persistent issue that intensifies during election season, as bad actors spread conspiracy theories and misinformation to manipulate voters. As the U.S. election approaches its conclusion in one of the closest races to date, researchers at Ben-Gurion University of the Negev have developed a method to assist fact-checkers in managing the growing amounts of misinformation on social media. Their findings were published recently in Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. The team, led by Dr. Nir Grinberg and Prof. Rami Puzis, discovered that monitoring fake news sources instead of individual articles or posts can greatly reduce the workload for fact-checkers and yield reliable results over time. Dr. Grinberg explains, “The problem today with the proliferation of fake news is that fact checkers are overwhelmed. They cannot fact-check everything, but the breadth of their coverage amidst a sea of social media content and user flags is unclear. “Moreover, we know little about how successful fact-checkers are in getting to the most important content to fact-check,” Dr. Grinberg added. “That prompted us to develop a machine learning approach that can help fact-checkers direct their attention better and boost their productivity.” Fake news sources often emerge and fade away rapidly over the years, making it challenging and resource-intensive to maintain updated lists of these sites. Their system considers the dynamics of information flow on social media and the audience’s willingness to engage with falsehoods. This approach allows for the identification of more sites and ensures greater resilience over time. The audience-based models developed by researchers significantly surpassed traditional methods for identifying misinformation, showing a 33% improvement when analyzing historical data and a 69% improvement with emerging sources over time. The authors also show that their approach can maintain the same level of accuracy in identifying fake news sources while requiring less than a quarter of the fact-checking costs. The system needs more training in real-world scenarios, and it should never replace human fact-checkers. However, “it can greatly expand the coverage of today’s fact checkers,” says Dr. Grinberg, a Department of Software and Information Systems Engineering member. Prof. Puzis is also a member of the same department. |
