Kees van Deemter

Kees van Deemter is a Professor in the Department of Computing Science at the University of Aberdeen's School of Natural and Computing Sciences. With a research career spanning over three decades, he has established himself as a leading figure in Natural Language Generation and Computational Linguistics. His work bridges theoretical linguistics with practical NLP applications, with particular focus on referring expression generation, multilingual processing, and evaluation methodologies.

Van Deemter's research interests center on Natural Language Generation, with special emphasis on Referring Expression Generation (REG), computational modeling of linguistic phenomena across languages, and the development of robust evaluation frameworks for NLP systems. His work spans theoretical foundations to practical applications, including logic-to-text generation, cross-linguistic studies comparing East Asian and European languages, and addressing challenges in hallucination and reproducibility in NLP.

His recent publications demonstrate a consistent focus on improving evaluation methodologies in NLP, with particular attention to task-based assessments rather than purely ratings-based approaches. He has made significant contributions to understanding how linguistic features influence reference production across languages, with extensive work on Chinese and Mandarin processing. His research group has developed several important corpora including QTUNA for quantifier studies and Mtuna for Mandarin referring expressions.

• Co-authored the influential survey 'Computational Generation of Referring Expressions: A Survey' (2012)
• Developed the Mtuna corpus for Mandarin referring expressions
• Contributed foundational work on the incremental algorithm for referring expression generation
• Pioneered research on vague language generation from a game-theoretic perspective
• Investigated cross-linguistic differences in reference production between English and East Asian languages

Van Deemter has supervised numerous PhD students who have become active researchers in NLP, including Guanyi Chen, Fahime Same, and Albert Gatt. His collaborative work extends across multiple institutions and has significantly shaped current approaches to referring expression generation and evaluation in NLP. His recent work critically examines the limitations of current evaluation practices in NLP and proposes more scientifically grounded assessment frameworks.