SPAIS1.7_Analysis_Share.xlsx

Geospatial Location Embedding (GLE) helps a Large Language Model (LLM) assimilate and analyze spatial data. GLE emergence in Geospatial Artificial Intelligence (GeoAI) is precipitated by the need for deeper geospatial awareness and the success of LLMs in extracting deep meaning in Generative AI. We searched Google Scholar, Science Direct, and arXiv for papers on geospatial location embedding and LLM and reviewed articles focused on gaining deeper spatial "‘knowing"’. We screened 304 titles, 30 abstracts, and 18 full-text papers that reveal four GLE themes - Entity Location Embedding (ELE), Document Location Embedding (DLE), Sequence Location Embedding (SLE), and Token Location Embedding (TLE). Synthesis is narrative, including a dialogic conversation between "‘Space"’ and "‘LLM.."’ GLEs themes reveal deeper spatial understanding comes via geospatial superimposing and show opportunities to advance in spatial modalities and generalized reasoning. Ultimately, GLEs signal the need for a <i>Spatial Foundation/Language Model</i>  (SLM) that embeds spatial knowing within the model architecture. The SLM framework advances Spatial Artificial Intelligence Systems (SPAIS) by establishing a Spatial Vector Space (SVS), mapping to physical space. SVS uniquely imbues an AI-capable space with a representation of actual space, paving the way for AI native geo storage, analysis, and multi-modality as the basis for SPAIS.