11-05, 14:00–14:30 (America/New_York), Lake Anne
Geconnex.us supports the creation of an open, community-contributed knowledge graph linking hydrologic features in the US to streamline water data sharing. The knowledge graph can answer innumerable water-related questions with an authority not yet standard in LLMs.
With the rapid growth of AI-powered large language models (LLMs), the internet is increasingly able to answer a broad range of questions. However, the reliability of these answers are often in doubt because they are not required to be truthful. Minor inaccuracies may be acceptable when drafting emails or summarizing articles, but they become far more consequential when the questions relate to public trust and safety. For AI systems to provide reliable, precise, and contextualized answers about the real world, they must be grounded in structured, authoritative data. Knowledge graphs offer a framework for linking diverse datasets across institutional and disciplinary boundaries, allowing AI models to retrieve more trustworthy answers.
The geoconnex.us project supports the creation of an open, community-contributed knowledge graph linking hydrologic features in the United States as an implementation of the Internet of Water Principles, particularly that “modern data infrastructure increases the usefulness of water data and enables its broadest possible application.” It offers persistent, canonical identifiers of real-world water-related entities, enabling data publishers to take full advantage of linked data while empowering data consumers to discover and reuse it. The project also drives contribution to the growing open source ecosystem of tools to publish and consume geospatial information products.
As an implementation of W3C Web Best Practices and the OGC Second Environmental Linked Features Interoperability Experiment, geoconnex.us extends standards like Google Structured Data Markup to support structured, semantic representations of rivers, lakes, watersheds, monitoring stations, and more. Web Pages that adopt this markup can be crawled and indexed into the public knowledge graph, thereby enabling authoritative, fact-driven answers.
This session will walk participants through the open source technologies that make this possible, how to contribute data, and how to access and navigate the graph, as well as demonstrating real-world use cases. It will explore how LLMs can interact with the geoconnex.us system to surface reliable, structured information about water in the United States and enable data-informed decision-making in water management.
Ben Webb is a software developer with the Internet of Water (IoW) project at the Lincoln Institute’s Center for Geospatial Solutions. Ben is working to develop core CGS and IoW software for water data management exchange to support state and federal agencies, as well as nonprofit organizations, addressing key climate resilience, conservation, and water management outcomes. A graduate of Colby College with a B.A. in computational biology, he is used to seeking answers to complicated questions by developing software and common data standards.
