Journalistic Inquiry On Public Debates

Photo of Joris Veerbeek

The rise of deep learning, and in particular that of the contextual language models named after Muppets, such as BERT, ELMo, and ERNIE, has led to significant breakthroughs in the field of Natural Language Processing (NLP). These models are able to process language much more accurately than their predecessors and in many
applications do not need to be trained completely from scratch.

Together with Dutch-language journalistic weekly De Groene Amsterdammer, the aim of this project is to investigate how these state-of-the-art language models can be fruitfully applied for journalistic inquiry, not only to offer better (online) journalistic products, but also for investigative journalism. In particular we investigate hate-speech and stance detection, and contextual embeddings and language representation.


Joris Veerbeek

Academic supervisors

Karin van Es, Mirko Tobias Schäfer, Kees van Deemter, Ayoub Bagheri, Beatrice de Graaf, Pim Huijnen

Grant funding agency and (co-)funding non-academic partners

De Groene Amsterdammer