Bridging the Gap: Enhancing Geospatial Analysis with Natural Language and Scenario Generation Language

Abstract

Scenario Generation Language (SGL) is a powerful tool that simplifies geospatial analysis and decision-making processes, removing the requirement for users to have expertise in GIS or SQL. However, users still need to understand the SGL grammar. This paper introduces a novel approach that utilizes GPT (Generative Pre-trained Transformer) - LLM (Large Language Model) to generate SGL statements directly from natural language questions. By leveraging the capabilities of GPT-LLM, this approach bridges the gap between user intent and technical query construction, enhancing the usability and accessibility of SGL. It enables decision-makers to interact with geospatial data using familiar natural language queries, without the need for in-depth knowledge of SGL or complex geospatial querying techniques. The integration of natural language processing with SGL empowers users to effortlessly generate accurate and syntactically correct statements, streamlining the analysis process and facilitating scenario exploration. Experimental results indicate that directly utilizing GPT-LLM for geospatial analysis may not yield satisfactory results. However, the approach presented in this paper demonstrates its effectiveness in simplifying geospatial analysis and supporting informed decision-making.