From Québec’s Assessment Reports to the SIGEOM GIS with a Human-in-the-Loop AI

The presentation discusses the combined initiative of the Ministry of Natural Resources and Forests (MRNF) in Québec and AgileDD to improve the accessibility of subsurface data for mining investors through the AgileDD AI platform. It addresses the challenges posed by unstructured assessment reports, which complicate data publication via GIS. The MRNF captures and segments sample and assay tables, along with relevant metadata, while ensuring data quality necessary for user trust.

It highlights a hybrid approach combining AI and technician oversight to efficiently process around 7,000 assessment reports annually, a task that would otherwise require over 100 technicians due to time-intensive manual data capture. Key strategies include utilizing multiple OCR systems for better accuracy, allowing end-user edits, and employing machine learning for continuous improvement. The integration of a Human-In-The-Loop approach enhances productivity, enabling a small team to generate 60,000 geolocated samples monthly. The paper concludes that while current AI technologies are not yet fully reliable for comprehensive geochemical data extraction, the partnership between AI and human expertise presents a promising solution.