Katalog Plus
Bibliothek der Frankfurt UAS
Bald neuer Katalog: sichern Sie sich schon vorab Ihre persönlichen Merklisten im Nutzerkonto: Anleitung.
Dieses Ergebnis aus BASE kann Gästen nicht angezeigt werden.  Login für vollen Zugriff.

Emergent Spatio-Semantic Structure in Large Language Model Embedding Spaces

Title: Emergent Spatio-Semantic Structure in Large Language Model Embedding Spaces
Authors: Shingleton, Joseph; Bicakci, Yunus Serhat; Wang, Yu; Basiri, Ana
Publisher Information: California Digital Library (CDL)
Publication Year: 2026
Description: Large Language Models (LLMs) are increasingly used in geospatial applications typically as generators of geographic text or as natural language interfaces to spatial data. Here, we explore whether LLM embedding spaces can instead function as geospatial representations that can be exploited directly. Using embeddings extracted from Airbnb property descriptions in London, we show that off-the-shelf LLM embeddings exhibit emergent spatial structure. We further demonstrate that a lightweight residual geo-adapter substantially sharpens this spatial signal, enabling approximate localisation even when explicit geographic references are removed, while preserving semantic relationships learned during LLM pre-training. These results suggest a path toward spatially explicit foundation models which operate over the spatio-semantic embedding space, rather than generated text.
Document Type: other/unknown material
Language: unknown
DOI: 10.31223/x5fq93
Availability: https://doi.org/10.31223/x5fq93; https://eartharxiv.org/repository/object/11894/download/21388/
Rights: https://creativecommons.org/licenses/by/4.0/legalcode
Accession Number: edsbas.826B1450
Database: BASE