Tom Kobes joined the project as a PhD student on 3 June 2026. He holds an MA in Digital Humanities and Digital Knowledge (University of Bologna) and a BA in History (University of Amsterdam). His Master’s thesis involved building a multi-agent NLP pipeline producing a knowledge graph mapping inter-document references across a historical archive, making claims explicit, traceable and anchored to evidence- experience he hopes to build on in the modelling and annotation challenges of this project.

Tom will be responsible for the following tasks:

• conduct a user study investigating how GIS experts conceptually interpret and use provenance information to answer geo-analytical questions;
• investigate and model the concepts needed to describe geodata provenance, building on existing models of core concepts and measurements of environmental phenomena;
• develop an NLP model to identify keywords that describes the content of a geodata source in terms of domains of measurement and geographic phenomena;
• develop and test a conceptual framework for annotating geodata sources with transformation graphs and content keywords;
• train and evaluate a machine learning model to automatically scale these annotations across a full geodata repository.