Mineral exploration requires the integration of heterogeneous geochemical, geophysical and geological evidence. Existing prospectivity tools often deliver opaque scores or heatmaps that are hard to interpret. MineTRACE is a web‑based interactive platform that supports evidence‑grounded exploration for eight commodities: Cu, Au, Ni, W, Sn, Co, Ta and Mn. Users can browse prospectivity maps, query specific locations or regions, inspect supporting evidence, and interact via natural language. The core of the system is a transparent expert tree that combines geological knowledge and known deposits to produce interpretable prospectivity scores from multi‑source evidence. For a new location, the conversational assistant retrieves the score and its evidence from the analysis pipeline and presents them in natural language. The scorer achieves spatial AUC values up to 0.917 across various test scenarios, and end‑to‑end evaluation measures query accuracy and grounding of responses. MineTRACE makes public geoscience data easier to access, interpret and verify, enabling more efficient and transparent mineral exploration.
Review: By coupling an explainable expert tree with natural‑language interaction, the system lowers the expertise barrier and offers trustworthy guidance for mining stakeholders.