Broadening Access to Transportation Safety Data with Generative AI: A Schema-Grounded Framework for Spatial Natural Language Queries
Researchers have developed a new framework that uses generative AI to make transportation safety data more accessible. This system translates natural language queries into structured operations, ensuring reproducible and schema-grounded results from a PostGIS database. An evaluation using Massachusetts transportation data showed that the validation layer corrected errors in 29% of queries, highlighting the challenge of aligning flexible language with strict data requirements. The approach aims to broaden access to critical safety information for public-sector planning. AI
IMPACT Enables broader access to critical safety data for public sector planning through natural language interfaces.