Transforming geospatial textual data into narrative storytelling visualization

Abstract

Current large language models (LLMs) often struggle to integrate geospatial data into dynamic, interactive visualizations, relying instead on text-based outputs. This limitation hinders the full potential of geospatial data to convey complex information through narrativedriven communication, making it difficult for users to interpret the data easily. Meanwhile, existing data visualization tools typically depend on static dashboards and rigid scientific formats, which have a steep learning curve and lack engagement through narrative elements. Audiences, however, are increasingly drawn to story-driven presentations, as seen in platforms pioneered by the MIT Senseable City Lab, and widely popularized by The New York Times and the Washington Post, which use narrative data visualization formats to attract and immerse readers. This gap between the capabilities of current LLM-based tools and users’ preferences presents a unique opportunity to develop a narrative-based geospatial visualization tool that meets these needs. This tool could transform how we communicate spatial data, particularly in fields such as journalism, travel planning, and urban planning, where the ability to convey complex patterns in an engaging manner is essential.