NGF AiNed Fellowship for democratizing insights from structured data

Institutions lacking resources are struggling to make use of huge amounts of structured data, creating a data literacy gap and making them fall behind large corporations. Madelon Hulsebos, an incoming tenure-track researcher, received an NGF AiNed Fellowship grant from the Dutch Research Council (NWO) for a project that enables anyone to make independent, data-driven decisions based on structured data.

Publication date
21 Mar 2024

By 2023, approximately 120 zettabytes of data has been collected worldwide, an exceptionally large quantity. To put this into perspective: if you were to stream HD movies, it's estimated that 1 zettabyte would allow you to stream movies continuously for many millions of years. But less than 1% of the 120 zettabytes is actually used. A large portion of this data is of a structured nature, for example tables, spreadsheets, and relational databases, which typically drive important decision-making processes in healthcare, governments and finance. Smaller companies, non-profit organisations, and public institutions are falling behind in developing their data analytics capabilities, creating an inequality in data literacy compared to large corporations.

The project

Artificial Intelligence (AI) could help, as it has proven to be highly effective in applications involving unstructured data (such as text) and images. However, proportional progress on structured data is currently lacking. With the DataLibra project, Hulsebos and her colleagues want to diminish these gaps by democratizing insight retrieval from (semi-)structured data through Table Representation Learning.

The goal is to provide trustworthy, secure, and responsible data analytics, empowering everyone to make data-driven decisions, easily, effectively and efficiently. The DataLibra project aims to tackle challenges throughout the entire data analytics pipeline. This includes efficient data storage and query execution, automated responsible data quality improvement, multimodal data integration and querying, and retrieval systems. The 5-year project will involve collaborations across various knowledge institutes and innovation labs due to its multidisciplinary nature.

About the fellowship

Hulsebos is one of five researchers who received a National Growth Fund AiNed Fellowship Grant. With this grants programme NWO aims to attract AI talent to Dutch academic research organizations. The programme promotes the development and application of artificial intelligence (AI) in Dutch businesses and governments, and was developed by the Netherlands AI Coalition. It started in 2022 and has awarded 14 grants in total. The DataLibra project receives over 900,000 euros from NWO.

Portrait of Madelon Hulsebos

About Madelon Hulsebos

Madelon Hulsebos is a postdoctoral fellow at UC Berkeley, and starts as a tenure track researcher at the Database Architectures group of CWI by fall 2024. She obtained her PhD from the Informatics Institute at the University of Amsterdam, for which she did research at MIT and Sigma Computing. Her research focuses on the intersection of data management and machine learning. She has recently made contributions to methods, tools, and resources for Table Representation Learning. She was awarded the BIDS-Accenture fellowship for her postdoctoral research on retrieval systems for structured data.

Image at top of this page is generated by AI (replicate.com)