Online harms

Risks online come from many sources: misinformation, propaganda, bullying, abuse. We research data and methods for detecting and analysing harms.

Efficient machine learning

Processing text doesn’t have to demand vast amounts of energy, especially with today’s tech. We research fast and green methods for NLP & machine learning.

Clinical NLP

Our research into clinical text processing covers both automatic reading of clinical journals and text processing to support public health and diagnoses.

NLP for Danish

Danish users should have access to modern language technology. We make our contribution to advancing Danish NLP research.

Generation

Generating language is complex, and processing the structures involved requires powerful, creative models. We research learning to generate sequences from data.