Anke Stoll, a member of the Computational Communication Sciece research group at TU Ilmenau, has published an article on the challenges of automated hate speech detection using machine learning. In her paper The accuracy trap or How to build a phony classifier, she describes typical pitfalls in the development and evaluation of machine-learning-based methods for automated classification of hate speech. As online participation increases, such methods are increasingly being used to automatically detect hate speech and then respond to it more quickly, e.g. by moderating, reporting or deleting it. Machine learning is also increasingly used in communication science to automatically analyze large amounts of research data such as user comments or journalistic articles.

The article appeared in the anthology Challenges and perspectives of hate speech research (Strippel et al., 2023) and can be accessed here: https: //

You can access the full anthology here: