Mirella Lapata
About
I am currently a professor in computer science at the the School of Informatics at the University of Edinburgh and a member of EdinburghNLP, the Natural Language Processing Group at the University of Edinburgh. My overarching research goal is to develop AI systems that not only follow patterns but reason, generalize, and adapt to novel situations. I aim to create models capable of understanding requests, aggregating and conveying information across modalities, forming long-term plans, and reasoning creatively about new challenges. My research interests focus on improving compositional generalization in deep learning models, enhancing cross-lingual transfer in multilingual language models, addressing limitations in understanding and generating long contexts, and creating verifiable systems, which generate responses with supporting evidence. Problems that I'm currently excited about include: (1) Agent-based frameworks for collaborative writing tasks (e.g., writing a story or a book chapter). (2) Improving the representation of long context (e.g., for movie summarization, video QA). (3) Parameter efficient approaches for LLM generalization to new tasks. (4) Language grounding (e.g., semantic parsing, multimodal representations).
Research Performance Summary
First Recorded Paper
Expanding the domain of a multi-lingual speech-to-speech translation system
Year: 1997
Citations: 20
Venue: Spoken Language Translation
Latest Recorded Paper
MT-PingEval: Evaluating Multi-Turn Collaboration with Private Information Games
Year: 2026
Citations: 0
Venue: arXiv preprint arXiv:2602.24188
Last 10 Years Publication Activity
This timeline shows the professor's yearly publication activity.