2026-08-30 –, 605
Attendees will build two complementary approaches for enabling geospatial awareness in LLMs: Geo-GraphRAGs for analyzing networks of features and DGGS-based methods for integrating aggregate and statistical data, and how they work together.
LLMs lack intrinsic geospatial awareness and do not maintain structured knowledge of geographic features or their domain-specific relationships. As a result, they cannot reason about how disruptions propagate through interconnected spatial systems—such as transportation networks, health service accessibility, or school catchments—during events like floods, wildfires, or other natural hazards. While LLMs excel at synthesizing text, they lack an internal representation of spatial networks and semantic relationships required for geospatial reasoning.
In this hands-on workshop, participants will build a geospatially aware LLM application from the ground up using open-source tools. You will create a Spatial Knowledge Graph (SKG) using Apache Jena to model networks of geographic features and their relationships, and use it to develop a Geo-GraphRAG pipeline that enables LLMs to translate natural-language questions into GeoSPARQL queries—providing transparent, traceable answers grounded in real data.
You will also learn how Discrete Global Grid Systems (DGGS)—an emerging Open Geospatial Consortium (OGC) standard—enable integration of aggregate and statistical data across domains by aligning datasets to a common spatial reference. Using open-source DGGS tooling such as DGGAL, participants will build an LLM agent capable of generating both GeoSPARQL queries over feature networks and DGGS API calls using CQL2, and explore how semantic feature networks relate to DGGS zones.
By the end of the session, you will understand how graphs and grid systems work together to unlock geospatial reasoning in AI systems.
This workshop builds on our FOSS4G North America 2025 session, now extended with DGGS to demonstrate the next generation of GeoAI workflows.
Attendees will need a GitHub account, as the workshop will be developed in a GitHub Codespace.
What skills do participants require to have?:Users will need basic Python skills or experience with a structured programming language.
Nathan McEachen is the founder, CEO, and CTO of TerraFrame, which specializes in supporting ministries of health and national spatial data infrastructures by building geospatial knowledge infrastructures with open-source GIS, remote sensing, and interoperability solutions. He obtained his master’s degree in computer science from Colorado State University in the United States. He is academically published in the fields of software testing, model-driven engineering, disease intervention, and spatial information sciences. Nathan is involved with the Open Geospatial Consortium and HL7 to help align standards development, enabling more automated data integration across sectors to bring geospatial awareness to LLMs.