Understanding RAG & GeoRAG - Building smarter knowledge systems

On December 10th, the PoliRuralPlus webinar series continued with a session dedicated to Retrieval-Augmented Generation (RAG) and its spatial extension, GeoRAG. The webinar, led by Filip Leitner (CCSS), provided a comprehensive introduction to how these technologies enhance large language models (LLMs) with external knowledge and spatial awareness.

The session opened with a discussion of RAG — a method that strengthens AI’s factual accuracy by combining language models with vector databases that store information as embeddings. This approach allows AI systems to “look up” relevant knowledge instead of relying solely on what they were trained on, mitigating common issues like hallucination and data staleness.

The presentation then introduced GeoRAG, a novel extension designed to make AI systems spatially aware. Filip demonstrated how GeoRAG enables natural language queries over geospatial datasets, linking semantic meaning with real-world locations. Participants learned how vector embeddings and metadata classification empower GeoRAG to interpret questions such as “Which areas are best suited for renewable energy installations?” or “Where do grasslands occur on gentle slopes?”

Three main query types were discussed:

• Semantic matching, connecting queries to datasets based on descriptions and metadata.
  
• Structured filtering, mapping user questions directly to database fields.
  
• Similarity-based querying, using vector representations to find related features or locations.
  

The team emphasized the importance of metadata and semantic enrichment for maximizing GeoRAG’s capabilities and discussed how it complements other PoliRuralPlus tools for regional analysis and planning.

The session concluded with an open Q&A, highlighting opportunities for collaboration and integration of RAG and GeoRAG technologies in future projects.

👉 Follow the PoliRuralPlus Tools Webinar Series through our website and social media channels to explore more on how AI and geospatial intelligence are transforming regional innovation and policy-making.