Keynote Speakers

This year we are happy to welcome two keynote speakers (one TBA) who are working on the cutting edge of spatial data science.

Live Interview

The symposium will feature a live/on-stage, interactive interview with Michael Goodchild from UC Santa Barbara.

Accepted Papers

This year 8 papers were accepted for presentation and publication in the SDSS 2025 proceedings.

Emerging Researcher Panel

This year's symposium will feature an Emerging Researcher Panel highlighting diverse voices from across the spatial data science community. Panelists at different career stages and from a range of disciplinary backgrounds will share their experiences, insights, and lessons learned as they navigated their own research paths. The session is designed to support new and early-career researchers by offering practical advice, candid reflections, and perspectives on building a meaningful and impactful career in spatial data science.
When is this session?

Generative AI and Agentic AI for GIS and Cartography

The session will explore the convergence of the latest exploration of Generative AI and Agentic AI in GIScience and Cartography, highlighting their transformative potential for geospatial data retrieval, analysis, and decision support. Generative AI, contributes powerful capabilities for creating synthetic spatial data, generating maps, coding, and simulating scenarios, offering new ways to support cartography and spatial data retrieval. Agentic AI, on the other hand, represents a major advance with its capacity for task planning, execution, adaptability, and goal-oriented autonomy in dynamic environments. By combining agentic coordination with generative and adaptive reasoning, these methods enable richer spatial representations, enhanced spatial decision-making, and more intelligent GIS workflows.
When is this session?

Building the Future of Open Mobility Data: Towards a Collaborative MobilityBank Platform

This session will examine the vision for a global, open, and privacy-preserving platform for GPS tracking studies, drawing on the recently proposed MobilityBank initiative (formerly OpenGPS). It will begin with two short presentations: one outlining the current landscape of mobility datasets, their limitations, and gaps, and another introducing the conceptual framework for MobilityBank as a centralized repository. Participants will then take part in breakout discussions to consider the design of such a platform, the technical, ethical, and institutional challenges it raises, questions of governance, and possible incentives for researcher participation. The session will conclude with a collective mapping of ideas and priorities to inform the next steps for MobilityBank.
When is this session?

Trustworthy (Geo)AI for Population Health

This Presentation Series will highlight cutting-edge GeoAI approaches in public health, focusing on multimodal, spatially explicit methods (e.g., imagery, mobility trajectories, surveys, and health data) that move from prediction to decision support. It will showcase advances in AI and vision–language models for geospatial health (e.g., exposure assessment, environmental epidemiology, mental/behavioural health), with attention to links between methodological innovation and population outcomes such as health equity and disease burden reduction. The series will also address robustness, fairness, and privacy-preserving techniques in real-world deployments, explore causal inference and counterfactual analysis to connect GeoAI outputs with public-health interventions, and share reproducible pipelines and standards for translating analytics into policy and practice.
When is this session?

GeoAI – Reflections on uncertainty and bias

Uncertainty and bias in GeoAI are central concerns for the GIScience community. This session brings together conceptual and applied perspectives, examining how large language models construct and reproduce geographic knowledge, often privileging certain regions or misrepresenting local contexts. Building on recent evaluations of tasks like geocoding and elevation estimation, we raise critical questions about treating LLMs as geographic knowledge generators and strategies for mitigating bias. Complementing this, research on uncertainty highlights diverse domains—such as wildfire risk, cattle movement, energy modeling, and supply-chain prediction—illustrating how measurement errors, sensor noise, algorithmic limitations, and incomplete knowledge shape outcomes. By distinguishing between aleatoric and epistemic uncertainty and discussing mitigation strategies like probabilistic modeling, ensemble learning, and uncertainty mapping, the session invites an interactive dialogue to shape a shared research agenda on equitable and trustworthy GeoAI.
When is this session?

Location Privacy in the Digital Society

As digital mapping and location-based services become integral to daily life, the issue of location privacy grows increasingly critical. This panel session will build upon the momentum from last year and continue exploring the intersection of privacy, technology, and geography, focusing on the ethical and technological challenges associated with geospatial data collection and usage. Participants will gain insights from experts on public perceptions of location privacy and innovative solutions on privacy protections.
When is this session?

Urban Morphology with Python: City Structure as a Predictor and a Target

This tutorial introduces urban morphometrics as a way to capture and analyze the built-up structure of cities, offering insights into their function, population, and development. Focusing on building footprints and street networks, it presents the momepy toolkit within the PySAL ecosystem, alongside GeoPandas and networkX, to demonstrate how spatial units can be generated and used to measure aspects of urban form such as size, shape, distribution, and configuration. These metrics will then be applied in two ways: to explain urban phenomena through spatial regression, and to detect distinct spatial patterns and hierarchies using classification models. Attendees will be able to code along on their machines or in-browser with Google Colab, and while only basic Python knowledge is expected, the principles introduced will be broadly transferable to other contexts.
When is this session?

GeoAI and Spatial Data Science for Disaster Research

Natural disasters, such as wildfires, structural fires, hurricanes, winter storms, floods, and earthquakes, pose significant threats to people and society. With climate change, these disasters are likely to become even more frequent and more costly in the near future. The availability of various geospatial data (e.g., satellite images, weather and climate data, mobile phone location data, historical disaster data, census data) combined with geospatial artificial intelligence methods provide great opportunities for understanding human-environment interactions under these disaster contexts. This session aims to bring together researchers interested in using GeoAI and spatial data science to answer disaster-related research questions and address issues in any aspect of disaster management, from mitigation and preparedness to response and recovery. Given the devastating impacts of recent wildfires and urban fire incidents, this session will also feature research on modeling fire risks and planning the spatial layout of fire services from a geospatial perspective.
When is this session?

Detecting and Evaluating AI Misalignment in Geographic Contexts

The problem of AI alignment concerns ensuring AI behaves in line with human values and goals. In geography, misaligned AI can distort spatial data, misguide urban planning, or reinforce mapping biases, with harmful consequences for communities and environments. This session offers a tutorial on detecting and evaluating AI misalignment in geographic contexts, focusing on (1) measuring geographic bias in AI performance, (2) the limited geographic diversity of large language models (LLMs), and (3) the misalignment of LLMs in delineating vague cognitive regions. We will demonstrate three projects: GeoBS, a Python toolbox for evaluating geographic bias; Prototype Explorer, an information observatory for analyzing geographic diversity and prototypes; and case studies on how LLMs define vague regions such as the Alps across scales and languages. Each co-organizer will lead a 15-minute sub-session, with 5 minutes reserved for Q&A.
When is this session?

mGeoHealth: real-world applications of spatial data science to improve population health and wellbeing.

This themetic session invites submissions in the area of mGeoHealth (Campbell et al., 2021; Campbell & Marek, 2022), which sits at the intersection of mobile health (mHealth) and GeoHealth (Health and Medical Geography) and often draws on spatial data science methods combined with mobile phone location data. The session will provide a forum for spatial data science researchers to consider applications in health and wellbeing, while also engaging health-focused scholars with quantitative spatial approaches. We particularly welcome submissions on social and physical environmental interactions with health outcomes (e.g., social connections, greenspace, mental health, happiness). We aim to attract 4–5 papers, with the possibility of lightning talks if submissions are high, and plan to propose a special issue on mGeoHealth in a journal such as Health & Place. Our goal is to strengthen engagement between spatial data science and health geography, advancing geographic data science for public good.
When is this session?

Organizing Committee

Symposium Hubs

SDSS 2025 is an online and distributed symposium. This means that participants are encouraged to join one of the symposium hubs distributed around the world. Groups of participants will meet at these hubs to present and discuss with other participants both in person and online. Joining a hub is not necessary, but encouraged.

To join a hub, email the contact listed for the city.

If you are interested in hosting a hub in your city, please complete this form.