Chat2Map: A ReAct-based Agent Framework for Automated Web Map Generation from Natural Language Instructions
Feb 16, 2026·
,,,,·
1 min read
Hongping Zhang
Peilong Ma
Cong Wang
Lei Ding
Zhen Wang
Heng Li
Abstract
This accepted full paper presents Chat2Map, a ReAct-based agent framework for automatically generating web maps from natural language instructions. It combines a layered toolset, fine-tuned code generation, geospatial data ingestion, and reliability constraints to improve executability and task completion for WebGIS map creation.
Type
Publication
ISPRS 2026 full paper accepted for oral presentation; to appear in the ISPRS Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences
This accepted conference paper introduces Chat2Map, a ReAct-based agent framework for translating natural language instructions into executable web map generation workflows with stronger reliability for WebGIS tasks.
