Online Library

The 8th Euromed Survey conducted by the European Institute of the Mediterranean touches
upon a number of important and complex issues related to violent extremism in the EuroMediterranean
region, including the question of the context and drivers through which violent
extremism can prosper. Echoing some of the results, this article looks into propaganda as a
tool of extremist ideologies and how to counter it.

VOX-Pol Blog post.

Lessons learned after one year of the NetzDG law.

In 2020, the Counter Extremism Project (CEP) Berlin carried out a sample analysis to test the extent to which YouTube, Facebook and Instagram block “manifestly illegal” terrorist content upon notification. Our findings raise doubts that the currently applied compliance systems of these companies achieve the objectives of making their services safe for users. As a result, CEP proposed requirements for effective and transparent compliance regimes with a focus on automated decision-making tools and lessons learned from the regulation of the financial industry. This Policy Paper combines all the relevant lessons learned and gives concrete recommendations on how to make the DSA also an effective regulation against terrorist content online.

VOX-Pol Blog post.

The increasing presence of hate speech on social media has drawn significant investment from governments, companies, and empirical research. Existing methods typically use a supervised text classification approach that depends on carefully engineered features. However, it is unclear if these features contribute equally to the performance of such methods. We conduct a feature selection analysis in such a task using Twitter as a case study, and show findings that challenge conventional perception of the importance of manual feature engineering: automatic feature selection can drastically reduce the carefully engineered features by over 90% and selects predominantly generic features often used by many other language related tasks; nevertheless, the resulting models perform better using automatically selected features than carefully crafted task-specific features.