~/about

Effy Xue Li.

Postdoctoral Researcher at CWI Amsterdam.
Working on Agentic Data Science, LLMs for Data Management and LLMs for Information Extraction.

Portrait of Effy Xue Li

Latest news

updated Oct 2026
show older news (12)
01 / Background

Hi! I’m Effy, you can also call me 李雪 [lǐ xuě]. I am a Postdoctoral Researcher at CWI Amsterdam, in the TRL lab, led by dr. ir. Madelon Hulsebos. I am interested in the topic of evaluating and building Agentic Data Science.

I did my PhD at University of Amsterdam in INDE lab, supervised by Prof. Paul Groth and Dr. Jan-Christoph Kalo. During my PhD, I did an internship at MotherDuck on data wrangling with LLMs. Previously I was an AI resident at Microsoft Research Cambridge, UK. Before that I did my Master’s at University of Edinburgh.

My PhD topic was on Knowledge Graph Construction from Conversational Data. Specifically, I look into how we can better utilize (large) language models to extract information such as entities and relations, and furthermore, make (large) language models more data-efficient, robust and adaptable. My PhD research was funded and situated in an NWO project IN-SIGHT.it.

I am originally from China and have lived in the UK and the Netherlands. When I am not in front of my laptop, you can often find me bouldering (at a complete beginner level), cycling or drinking coffee (like everyone else :D). If you are keen to collaborate or exchange ideas over a coffee, you can reach out to me at effy DOT li AT cwi DOT nl.

02 / Interests
  • Agentic Data Science
  • LLMs for Data Management
  • LLMs for Information Extraction
03 / Academia
  • 2025–now Postdoctoral Researcher
    CWI Amsterdam
    Agentic Data Science.
  • 2020–2025 Ph.D., Knowledge Graph Construction
    University of Amsterdam
    Supervised by Prof.Paul Groth and Dr.Jan-Christoph Kalo.
  • 2017–2018 M.Sc., Artificial Intelligence
    University of Edinburgh
    Supervised by Prof. Shay Cohen.
  • 2012–2016 B.Eng., Electronic Information Engineering
    Guangzhou University
    Graduated with top 1% GPA with National Scholarship.
04 / Industry
  • 2024 Intern
    MotherDuck
    Data wrangling with LLMs.
  • 2018–2019 AI Resident
    Microsoft Research Cambridge, UK
    API misuse detection with graph neural networks on source code; neural ranking for enterprise search.
  • 2016–2017 Front-end Engineer
    Guangzhou Wuliudian Education Consulting
    WeChat mini program, company homepage, UI and visual design.
05 / Publications
  • 2026 The Silent Failures in Agentic Data Science
    Xue Li, Hongqian Xia, Daniel Gomm, Aécio Santos, Cornelius Wolff, Madelon Hulsebos
    arXiv preprint [paper] [pdf] [project]
  • 2026 Beyond Scale and Generation: Understanding Language Model-based Entity Matching
    Zeyu Zhang, Xue Li, Iacer Calixto, Paul Groth, Sebastian Schelter
    arXiv preprint, under review [code] [paper] [pdf]
  • 2025 Towards a Controllable and Realistic Entity Matching Benchmark
    Xue Li, Zeyu Zhang, Sebastian Schelter, Paul Groth, Madelon Hulsebos
    DBDBD'25
  • 2025 Evaluation of Unsupervised Static Topic Models' Emergence Detection Ability
    Xue Li, Ciro D. Esposito, Paul Groth, Jonathan Sitruk, Balazs Szatmari, Nachoem Wijnberg
    PeerJ Computer Science [paper] [pdf]
  • 2025 Position: Unlocking the Full Potential of Data Science Requires Tabular Foundation Models, Agents, and Humans
    Tianji Cong, Julian Martin Eisenschlos, Daniel Gomm, Leo Grinsztajn, Andreas C. Mueller, Anupam Sanghi, Jan-Micha Bodensohn, Vadim Borisov, Michael Cochez, Katharina Eggensperger, Floris Geerts, Myung Jun Kim, Andreas Kipf, Xue Li, Olga Ovcharenko, Paolo Papotti, Lennart Purucker, Sebastian Schelter, Immanuel Trummer, Gaël Varoquaux, Liane Vogel, Carsten Binnig, Madelon Hulsebos, Frank Hutter
    Position paper, preprint [paper] [pdf]
  • 2024 How different is different? Systematically identifying distribution shifts and their impacts in NER datasets
    Xue Li, Paul Groth
    Language Resources and Evaluation (LREC) [paper]
  • 2024 Towards Efficient Data Wrangling with LLMs using Code Generation
    Xue Li, Till Döhmen
    DEEM@SIGMOD'24 [code] [pdf] [poster]
  • 2023 Knowledge-centric Prompt Composition for Knowledge Base Construction from Pre-trained Language Models
    Xue Li, Anthony Hughes, Majlinda Llugiqi, Fina Polat, Paul Groth, Fajar J. Ekaputra
    ISWC 2023 LM-KBC workshop [code] [pdf] [poster]
  • 2023 Do Instruction-tuned Large Language Models Help with Relation Extraction?
    Xue Li, Fina Polat, Paul Groth
    ISWC 2023 LM-KBC workshop [pdf]
  • 2021 The Challenges of Cross-Document Coreference Resolution for Email
    Xue Li, Sara Magliacane, Paul Groth
    Proceedings of the 11th Knowledge Capture Conference [pdf]
06 / Service
07 / Teaching
  • Jan – Jul 2026 Student supervision
    Student: Hongqian Xia
    Why do Data Science Agents fail?
  • Oct – Dec 2023 Teaching Assistant, NLP1, UvA
  • Jan – Jul 2023 Master thesis supervision
    Student: Casper Smit
    Graphical information for cross-document coreference resolution in emails [student]
  • Apr – May 2023 Teaching Assistant, Data and Knowledge, UvA
  • Jan – Jun 2022 Master thesis supervision
    Student: Timothy Dorr
    Not All Names Are WEIRD: Identifying Name Origin Bias in Named Entity Recognition Tasks on Email Data [student] [thesis]
  • Feb – Apr 2022 Teaching Assistant, Causal Data Science, UvA