Dr Stef De Sabbata

Stef De Sabbata is an Associate Professor of Geographical Information Science at the School of Geography Geology and the Environment, Senior Fellow at the Institute for Digital Culture, and Turing liaison (academic) of the University of Leicester. Her research focuses on geographical artificial intelligence (GeoAI), including the development of spatially-explicit approaches to urban analytics and the study and use of large language models, foundation models and generative artificial intelligence in geography and cultural analytics. One of her main current lines of research focuses on developing spatially-explicit approaches to urban analytics. In 2021, she organised a session at the Annual International Conference of the Royal Geographical Society with IBG, which sparked discussions and collaborations that finally led her to be the lead guest editor for a special issue of the International Journal of Geographical Information Science on GeoAI in Urban Analytics, including her paper on using graph neural networks in geodemographic classification. 

Stef De Sabbata is the Chair of the Geographic Information Science Research Group of the Royal Geographical Society with IBG. She is also part of the steering committee of GIScience Research UK (GISRUK), the chair of the GISRUK 2018 conference, and a member of the Commission on Location-Based Services of the International Cartographic Association.

Before joining the University of Leicester in 2015, Stef De Sabbata was a Researcher at the Oxford Internet Institute of the University of Oxford (2013-2015) and a Junior Research Fellow at the Wolfson College of the University of Oxford (2014-2015), and thereafter, a Research Associate of the Oxford Internet Institute of the University of Oxford (2015-2021). She was awarded a PhD from the Department of Geography of the University of Zurich in 2013 and a BSc and an MSc in computer science from the Department of Mathematics and Computer Science of the University of Udine.

I am a geographic data scientist working at the intersection between geography and artificial intelligence.

My current research focuses on developing spatially explicit machine learning approaches, particularly in urban analytics. In 2021, I organised a session at the Annual International Conference of the Royal Geographical Society with IBG. This session sparked discussions and collaborations that led me to guest edit a special issue of the International Journal of Geographical Information Science on GeoAI in Urban Analytics. This issue includes my paper on using graph neural networks for geodemographic classification and I have also applied graph neural networks and heterogeneous graph neural networks to study urban form. As part of my research, I am also interested in the use and interpretability of foundation models in geography and cultural analytics. My recent work on the geospatial mechanistic interpretability of large language models demonstrates how spatial analysis can be used to unveil how large language models "think" about geographic information.

My research also aims to leverage artificial intelligence to understand everyday geographies, the emerging meaning we attach to geographic places through the content generated on internet platforms. I collaborate with colleagues on topics that span through the wide range of research topics represented in the School from the geographies of everyday multicultural living in Leicester to the study of the Anthropocene, and the University from graph theory to cultural analytics. I am the co-I of the Museum Data Service, a free new service that aims to connect and share all the object records across all UK museum. I will be leading the outreach activities of the Museum Data Service and, in the past, I have collaborated with the Science Museum Group to develop new approaches to analysing oral history archives using large language models and visual analytics. 

De Sabbata, S., Mizzaro, S., & Roitero, K. (2025). Geospatial Mechanistic Interpretability of Large Language Models. arXiv preprint arXiv:2505.03368. [GitHub repo]

Liu, P., Wang, Y., De Sabbata, S., Lei, B., Biljecki, F., Tang, J., & Stouffs, R. (2025). Living upon networks: A heterogeneous graph neural embedding integrating waterway and street systems for urban form understanding. Environment and Planning B: Urban Analytics and City Science, 23998083251358527.

Sellers, H., Williams, M., Berrio, J. C., De Sabbata, S., Rose, N. L., Turner, S. D., ... & Aquino-Lopez, M. A. (2025). A mid-20th century stratigraphical Anthropocene is recognisable in the birth-area of the industrial revolution. The Anthropocene Review, 12(2), 176-200.

Bennett, K., De Sabbata, S., & Gardner, Z. (2025). Living together (and apart) on the move: new directions in everyday multiculturalism. Ethnic and Racial Studies, 1-21.

De Sabbata, S., Bennett, K., & Gardner, Z. (2024). Towards a study of everyday geographic information: Bringing the everyday into view. Environment and Planning B: Urban Analytics and City Science, 51(6), 1354-1368.

De Sabbata, S. and Liu, P. (2023). A graph neural network framework for spatial geodemographic classification. International Journal of Geographical Information Science, 37(12), pp. 2464–2486. [GitHub repo]

De Sabbata, S., Ballatore, A., Miller, H.J., et al. (2023). GeoAI in urban analytics. International Journal of Geographical Information Science, 37(12), 2455-2463.

De Sabbata, S., Ballatore, A., Liu, P., & Tate, N. (2023). Learning urban form through unsupervised graph-convolutional neural networks. The 2nd International Workshop on Geospatial Knowledge Graphs and GeoAI: Methods, Models, and Resources,12th September, 2023, Leeds, UK. [GitHub repo]

Bennett, K., Gardner, Z. and De Sabbata, S., 2023. Digital geographies of everyday multiculturalism: ‘Let’s go Nando’s!’. Social & Cultural Geography, 24(8), pp.1458-1477.

Bennett, K. and De Sabbata, S., 2023. Introducing a more-than-quantitative approach to explore emerging structures of feeling in the everyday. Emotion, Space and Society, 49, p.100965.

Gardner, Z., Bennett, K. and De Sabbata, S., 2023. Virtual reality, place and affect. A Research Agenda for Digital Geographies, p.69.

Ballatore, A., and De Sabbata, S. (2020). Los Angeles as a digital place: The geographies of user-generated content. Transactions in GIS, 24(4), 880-902. doi:10.1111/tgis.12600

Gardner, Z., Mooney, P., De Sabbata, S., and Dowthwaite, L. (2020). Quantifying gendered participation in OpenStreetMap: responding to theories of female (under) representation in crowdsourced mapping. GeoJournal, 85(6), 1603-1620. doi:10.1007/s10708-019-10035-z

De Sabbata, S., and Liu, P. (2019). Deep learning geodemographics with autoencoders and geographic convolution. In 22nd AGILE Conference on Geo-information Science.

Bright, J., De Sabbata, S., Lee, S., Ganesh, B., and Humphreys, D. K. (2018). OpenStreetMap data for alcohol research: Reliability assessment and quality indicators. Health & Place, 50, 130-136. doi:10.1016/j.healthplace.2018.01.009

Acheson, E., De Sabbata, S., and Purves, R. S. (2017). A quantitative analysis of global gazetteers: Patterns of coverage for common feature types. Computers, Environment and Urban Systems, 64, 309-320. doi:10.1016/j.compenvurbsys.2017.03.007

Reichenbacher, T., De Sabbata, S., Purves, R. S., and Fabrikant, S. I. (2016). Assessing geographic relevance for mobile search: A computational model and its validation via crowdsourcing. Journal of the Association for Information Science and Technology, 67(11), 2620-2634. doi:10.1002/asi.23625

Graham, M., De Sabbata, S., and Zook, M. A. (2015). Towards a study of information geographies: (im)mutable augmentations and a mapping of the geographies of information. Geo: Geography and Environment, 2 (1), 88-105. doi:10.1002/geo2.8

De Sabbata, S., Mizzaro, S., and Reichenbacher, T. (2015). Geographic dimensions of relevance. Journal of Documentation, 71(4), 650-666. doi:10.1108/JD-12-2013-0167

De Sabbata, S., and Reichenbacher, T. (2012). Criteria of geographic relevance: An experimental study. International Journal of Geographical Information Science, 26(8), 1495-1520. doi:10.1080/13658816.2011.639303

I have supervised two PhD students to completion and one to submission as the first supervisor. My first PhD student and now close collaborator, Dr Pengyuan Liu, recently accepted a lectureship in Digital Planning at the University of Glasgow. Dr Hannah Sellers recently completed a highly interdisciplinary project on resilient, biodiverse forests and has joined the Leicestershire and Rutland Wildlife Trust as Senior Nature Recovery Officer. Shahreen Nawfee has recently submitted her PhD dissertation focused on urban place representation through user-generated content. 

I am currently the first supervisor of a PhD student working on geospatial mechanistic interpretability, and I am also co-supervising PhD students based in geography, geology, computer science, criminology, museum studies, and media and communication. I have supervised over fifty students from taught degrees who have continued onwards to successful careers and doctoral studies, including students who won the departmental Best Master's Student Prize and Best GIScience Master's Dissertation Prize, and the national Best Dissertation Prize awarded by the GIScience Research Group of the Royal Geographical Society with IBG.

I am open to supervise PhD project focused on:

• Geospatial artificial intelligence (GeoAI)
• Geographic data science
• Geographic information science
• GIS
• Geocomputation
• Artificial intelligence
• Machine learning
• Deep learning
• Cultural analytics
• Natural language processing
• Information geographies
• Digital geographies
• Internet geographies
• Information visualisation
• Cartography

Data science is a crucial aspect of my teaching. I teach a number of modules that focus on applying data science approaches in geography. I am particularly excited about my GY7708 Geospatial Artificial Intelligence module where I have the opportunity to discuss a lot of my research. I also teach the GY3421 and GY7413 modules on Information Visualisation across a range of master programmes from the School of Geography, Geology and the Environment and the School of Computing and Mathematical Sciences, taking a systematic approach to visualising data based on the grammar of graphics and using R and Tableau.

In previous years, I taught a module covering the basics of programming in R and machine learning but it puts a significant emphasis on reproducibility which I think is a key aspect of data science. One of the ten teaching weeks focuses entirely on reproducibility and both pieces of coursework are expected to be reproducible analysis documents. Following the principle of teaching by example I created all materials (including lecture slides and practical session instructions) in R and Markdown and are thus fully human-readable reproducible using free and open-source software and accessible as bookdown via GitPages (and GitHub repo).

Geographic information science

Geographic data science

Geographic artificial intelligence

Geocomputation 

GIS

Location-based services

Geographic information retrieval

Natural language processing

Digital geographies 

Volunteered geographic information

Information visualisation

Cartography

Urban geography

Chair of the GIScience Research Group of the Royal Geographical Society

Member of the steering committee of GIScience Research UK (GISRUK)

Chair of the 26th Annual GIScience Research UK (GISRUK) Conference 2018

Member of the commission of the Commission on Location Based Services of the International Cartographic Association (ICA/ACI)

Journal of Documentation Highly Commended Paper Award 2016 for: De Sabbata S. Mizzaro S. & Reichenbacher T. (2015). Geographic dimensions of relevance. Journal of Documentation 71(4) 650-666.

Leicester Students’ Union Superstar Award nomination for teaching (2016-17) best personal tutor (2017-18 and 2020-21) and best supervisor (2020-21).

Fellow of the Royal Geographical Society (with IBG)

Fellow of the Higher Education Academy