ChatGeo: Forging Trust in Spatial Semantic Document and Knowledge Retrieval

The Challenge

Valuable data and knowledge contained within many traditional geoscience resources is under-utilised in exploration industry practice due to its unstructured nature. Large language models and large multimodal models offer promise in accessing such resources, but current AI technologies cannot interrogate spatially and contextually complex domain specific information in geological reporting.

Proposed Solution

This project aims to develop a geological-language-specific web portal to act as a trustworthy digital assistant supporting search and interrogation of geological resources including reports with multimodal information. Operating as a co-pilot through multi-turn natural language queries, this virtual assistant would help a user explore relevant digital assets and retrieve diverse information relevant to their needs, supporting efficient high-value knowledge extraction and synthesis.

Proposed Benefits to WA

By streamlining the extraction and analysis of geological information from varied resources, this project will enhance the competitive advantage of exploration companies working with the detailed pre-competitive and historical data available in Western Australia, supporting improved data utilisation and more robust evidence-based decision making in exploration planning and investment.