Wessel Poelman
I work as a PhD student at KU Leuven, where I am part of LAGoM under the supervision of Miryam de Lhoneux. My research asks a simple question: how do we know that multilingual NLP systems are improving for the reasons we think they are?
Before coming to Belgium, I worked at the Technical University of Munich as a research associate, completed a master’s in Information Science at the University of Groningen, and worked as a machine learning engineer at Web-IQ. More background is available on the About page or in my CV.
KU Leuven Profile · Google Scholar · Semantic Scholar · GitHub
Contact: wessel.poelman [at] kuleuven.be or contact [at] wesselpoelman.nl
Research
My work focuses on multilingual natural language processing: language sampling, tokenization, modeling choices, and evaluation. I am particularly interested in how linguistic differences between languages influence both model behavior and the claims we make about multilingual systems.
Current themes:
- Evaluation and experimentation methodology for multilingual NLP (1, 2, 3)
- Interaction of language characteristics and language modeling (4, 5, 6)
- Tooling for multilingual language technology (7, 8)
If this sounds interesting, feel free to reach out!
Selected Publications
* shared first authorship
For a complete list, see my publications page.
Recent News
- Jun 2026. Joining Barbara Plank at LMU for a three-month research stay.
- Apr 2026. Paper on multilingual Wikipedia data quality accepted at ACL.
- Mar 2026. Two papers at EACL.
- Dec 2025. Research visit in Copenhagen, with talks at Aalborg University and ITU.
- Nov 2025. Oral presentation at EMNLP.