Mugdha Pandya

Research keywords: Natural Language Processing, Bias and Fairness in NLP, Hate Speech Detection

Bio:

Mugdha’s research interests lie in addressing societal bias in NLP through cross-disciplinary approaches. She completed her MSc in Artificial Intelligence at the University of Edinburgh (2019–2020), building on a B.Tech in Computer Science from VIT, India (2014–2018). Her early research included multilingual speech processing and bias evaluation in hate speech detection.

She went on to work at Heriot-Watt University (2020–2021), where she investigated gender bias in conversational assistants, before joining the University of Sheffield as a Research Associate (2021–2025). At Sheffield, she worked on intersectional hate speech, minority opinion representation, and NLP applications in the digital humanities, collaborating with journalists and social scientists.

PhD research:

Current NLP systems often overlook the perspectives of minority groups, leading to blind spots in how harmful content is identified and addressed. Mugdha’s PhD research will explore how hate speech detection can be made more inclusive across cultural and linguistic boundaries.

She will examine how diverse understandings of harm can be integrated into dataset design, model training, and evaluation. By drawing on interdisciplinary methods, her work aims to develop language technologies that reflect a wider range of social and cultural contexts.

Supervisors: Björn Ross, Zeerak Talat