Automatic Detection of Manipulative Consent Management Platforms and the Journey into the Patterns of Darkness

Abstract

We study how to automatically classify different types of manipulative interface design pattern for Content Management Platforms (CMPs), also known as a cookie consents. Our approach uses a scraper to extract different features of CMPs. We then classify the CMP, based on these features, into one of five patterns defined specifically for CMPs. We evaluate our automatic "detector" using four different statistical measures. We also consider factors that cause misclassifications and discuss how to potentially avoid them.