DisCo – Disinformation and COVID-19

Misinformation and fake news are circulating worldwide regarding the war in Ukraine, the COVID-19 pandemic, and other current issues that are highly emotionally charged. People are sharing unverified news, particularly on social media platforms such as Facebook, Twitter, and YouTube, causing it to spread rapidly. “We are not only fighting a pandemic, but also an infodemic,” said Tedros Adhanom Ghebreyesus, Director-General of the WHO. Researchers at Fraunhofer SIT aim to address this challenge through the DisCo project.

Methods of Creating Fake News

One goal of the project is to first provide an overview of the current situation regarding fake news: What are the typical methods used to produce fake news in the context of a current conflict? What technical methods do fake news producers use? The researchers will then examine the latest technical solutions in the field of text and multimedia forensics and evaluate which of these methods are best suited to detecting fakes in texts, images, and videos. For example, the plan is to test the capabilities of reverse image searches from various search engines, which are often used to analyze suspicious image material. This allows users to identify images that have been taken out of their original context (without alteration).

Fact-checking tool for journalists

Another goal of DisCo is to use text forensic methods to develop a demonstrator for an annotation tool designed to help journalists and fact-checkers quickly assess and evaluate texts. The tool uses an ML algorithm trained by Fraunhofer SIT researchers as well as natural language processing (NLP) techniques. For fact-checkers and journalists, texts are formatted so that passages requiring verification are highlighted in color. Thus, texts are not automatically categorized as “fake” or “true”; instead, the text passages that the tool identifies as requiring verification are clearly highlighted. The journalist or fact-checker makes the final decision.