Dr. Fatma Elsafoury

In her research, Fatma Elsafoury is interested in social computing, in particular questions related to social bias, toxicity, hate speech and fairness. She is interested in understanding how social bias influences the decisions of ML models in terms of toxicity and hate speech detection. Her dissertation focuses on the effects of bias in NLP models on hate speech detection.

As part of her current research, she would like to extend the study of bias to non-Western perspectives and multilingual language models. She would also like to compare the sources of bias in NLP models with societal prejudices. In addition, she organizes the Women_in_NLP lecture series, to which she has so far invited 10 women from Google, Microsoft, Rasa, Allen AI, Carnegie Mellon University and others to talk about their research, share their experiences in NLP and encourage more women to work on NLP projects.