A novel data-knowledge dual-driven model coupling artificial intelligence with a mineral systems approach for mineral prospectivity mapping

Abstract Mineral prospectivity mapping (MPM) is recognized as an essential tool for targeting new mineral deposits. MPM typically comprises two end-member approaches: knowledge-driven and data-driven. Knowledge-driven MPM relies on expert knowledge, which is based on causal relationships but is not readily adaptable to dynamic changes. Data-driven MPM is capable of identifying underlying data patterns but involves poorly interpretable decision logic. Combining the advantages of knowledge-driven and data-driven paradigms is a research frontier in MPM. In this study, we designed a data-knowledge dual-driven model coupling artificial intelligence (AI) with a mineral systems approach to MPM. This model can utilize mineral systems as a guideline for data-driven AI to reasonably implement data selection, proxy extraction, and model operation for MPM. The newly developed data-knowledge dual-driven model achieved superior predictive performance and offered better interpretability compared to pure data-driven MPM.