The spread of fake news, hate speech, and manipulated content on social media is posing increasingly significant challenges for society, the media, and institutions. With the AI-powered analysis tool “DisDETECT,” computer scientist Dr. Mina Schütz, a researcher at the Center for Digital Safety & Security, has now developed a system that helps experts identify problematic content in a targeted manner and evaluate it more thoroughly.
As part of her dissertation at Darmstadt University of Applied Sciences, she developed a web-based tool at the Doctoral Center for Applied Computer Science (PZAI) that analyzes texts based on more than 50 characteristics. Unlike traditional fact-checking systems, “DisDETECT” does not automatically classify content as “true” or “false.” Instead, the system highlights conspicuous patterns and helps journalists, researchers, and experts from government agencies and organizations efficiently identify potentially critical content.
“The final decision always rests with humans,” emphasizes Mina Schütz. The system is intended to support the analysis, but not to make final decisions.
The tool analyzes large volumes of text and identifies, among other things, emotional language, indications of hate speech, disinformation, conspiracy theories, and typical clickbait patterns. The results are visualized interactively and presented using a traffic-light system with four color levels—ranging from trustworthy to critical. In the background, 19 AI models work together, and their results are combined to enable the most informed assessment possible. Additionally, the system displays a confidence interval, thereby making transparent the potential probability of error in the analysis.
Mina Schütz placed particular emphasis on the system’s human-centered approach. Using “visual analytics” methods, users can decide for themselves what information to display and how deeply they want to delve into the analysis. Initial tests with potential user groups have already yielded positive feedback. However, “DisDETECT” is currently still in the prototype stage; further development work is required before it can be widely implemented in practice.
The dissertation was a collaborative effort between Darmstadt University of Applied Sciences and the AIT Austrian Institute of Technology. Mina Schütz is now continuing her research at the AIT’s Center for Digital Safety & Security, where she is further expanding her work on AI-based analysis of disinformation and digital threats.
Further information: