Business Analytics Seminar: Madelon Hulsebos (CWI, UvA)
"Towards robust open-domain analytical querying over structured data"
Event details from: Business Analytics Seminar: Madelon Hulsebos (CWI, UvA)
- Date
- 14 July 2025
- Time
- 12:00 – 13:00
Abstract:
Structured data – such as relational tables and spreadsheets — remains a prevailing data modality in the organizational landscape, informing high-stake decisions. As a result, tabular reasoning has emerged as a key area of research. In this talk, I’ll introduce the task of open-domain querying over structured data, which unifies retrieval and generation to support natural language analytical queries without prior identification of relevant tables. I’ll present insights from our TARGET benchmark, with respect to effective table retrieval patterns, the role of metadata, and the potential of fine-grained query decomposition. Finally, I’ll present reflections on the robustness of tabular embedding models and LLMs for reasoning over relational data, raising important questions about how to evaluate and advance progress in this space.
Biography:
Madelon Hulsebos is a tenure track researcher at CWI in Amsterdam. Prior to that, she was a postdoctoral fellow at UC Berkeley, and obtained her PhD from the University of Amsterdam for which she did research at MIT and Sigma Computing. Her general research interest is on representation learning and generative models for tabular data to democratize insights from structured data. Madelon founded the Table Representation Learning workshop in 2022 at NeurIPS, and leads various other efforts in this space. She was awarded a BIDS-Accenture fellowship for her postdoctoral research at UC Berkeley as well as an AiNed fellowship grant funding her Table Representation Learning Lab for 5-year.
General information:
This seminar will be organised in a hybrid setup. If you are interested in joining this seminar, please send an email to the secretariat of Amsterdam Business School at secbs-abs@uva.nl.
