Geographic information system (GIS) users and analyst need to fetching geospatial data for analysis or research tasks. Data fetching can be time-consuming and label intensive. Our recent study proposes LLM-Find, an autonomous GIS agent framework to retrieve geospatial data by generating and executing programs with self-debugging. LLM-Find adopts an LLM as the decision maker to pick up the applicable data source from a list and then fetch data from the selected source. Each data source has a pre-defined handbook that records the metadata and technical details for data fetching. The proposed framework is flexible and extensible, designed as a plug-and-play mechanism; human users or autonomous data scrawlers can add a new data source by adding a new handbook. LLM-Find provides a fundamental agent framework for data fetching in autonomous GIS. We also prototyped an agent based on LLM-Find, which can fetch data from OpenStreetMap, COVID-19 cumulative cases from GitHub, administrative boundaries and demographic data from the US Census Bureau, weather data from a commercial provider, satellite basemap from ESRI World Imagery, and worldwid DEM from OpenTopography.org.
We tested various data cases; by accepting data requests in natural language, most of the requests got correct data in an about 80% – 90% success rate. We feel excited about that because the success of such data fetching agent indicates that the data intensive GIS research or boarder scientific research can be executed by agents. Autonomous research agents can collect necessary online or local data and then conduce analysis parallely while adjust methods or strategies for better results. LLM-Find will be a foundational role in such a bright vision.
For more details, please refer to our paper: Ning, Huan, Zhenlong Li, Temitope Akinboyewa, and M. Naser Lessani. 2024. “An Autonomous GIS Agent Framework for Geospatial Data Retrieval” arXiv. https://doi.org/10.48550/arXiv.2407.21024. GitHub repository: https://github.com/gladcolor/LLM-Find
Further reading: Autonomous GIS: the next-generation AI-powered GIS. Recommended citation format: Li Z., Ning H., 2023. Autonomous GIS: the next-generation AI-powered GIS. International Journal of Digital Earth. GitHub repository: https://github.com/gladcolor/LLM-Geo
