 |
VOOZH | about |
|
VOOZH | about |
| Reference Type | Journal (article/letter/editorial) | ||
|---|---|---|---|
| Title | Mineral prospectivity Mapping: An interpretable classifier Combining catchment basin and knowledge graph embedding | ||
| Journal | Ore Geology Reviews | ||
| Authors | Yan, Qun | Author | |
| Pei, Yao | Author | ||
| Xue, Linfu | Author | ||
| Sun, Hairui | Author | ||
| Wang, Rui | Author | ||
| Ran, Xiangjin | Author | ||
| Year | 2025 | Volume | < 184 > |
| Page(s) | 106758 | ||
| URL | |||
| DOI | doi:10.1016/j.oregeorev.2025.106758Search in ResearchGate | ||
| Generate Citation Formats | |||
| Classification | Not set | LoC | Not set |
| Mindat Ref. ID | 18621956 | Long-form Identifier | mindat:1:5:18621956:6 |
| GUID | 0 | ||
| Full Reference | Yan, Qun; Pei, Yao; Xue, Linfu; Sun, Hairui; Wang, Rui; Ran, Xiangjin (2025) Mineral prospectivity Mapping: An interpretable classifier Combining catchment basin and knowledge graph embedding. Ore Geology Reviews, 184. 106758 doi:10.1016/j.oregeorev.2025.106758 | ||
| Plain Text | Yan, Qun; Pei, Yao; Xue, Linfu; Sun, Hairui; Wang, Rui; Ran, Xiangjin (2025) Mineral prospectivity Mapping: An interpretable classifier Combining catchment basin and knowledge graph embedding. Ore Geology Reviews, 184. 106758 doi:10.1016/j.oregeorev.2025.106758 | ||
| In | Link this record to the correct parent record (if possible) | ||
| Abstract/Notes | The data-driven Mineral Prospectivity Mapping (MPM) lacks decision-making logic, whereas the knowledge-driven MPM relies on experience summarization. In this study, we develop an MPM method that integrates data and knowledge and conducts geological interpretation on the prediction model. We used the data derived from stream sediment geochemical assays, and the knowledge comprising the vectors embedded from a geological map knowledge graph. We construct catchment basins and use basin boundaries to constrain the spatial geochemical interpolation. A Graph Attention Network (GAT) is then employed to embed the geological map knowledge. The knowledge-embedded vectors are spatially mapped and aligned with the gridded geochemical data, and K-means clustering is applied to classify the samples spatially. Finally, these classified samples are used as the training dataset for the XGBoost model for gold prospectivity prediction. Shapley values are introduced to interpret the contribution of various data inputs. To validate the effectiveness of the methods, gold prospectivity prediction was conducted in the Mawu area (Gansu Province). The results demonstrate that by setting the spatial classification to 14 and integrating knowledge graph embedding, the AUC value of the XGBoost model increased by 3%, while the predicted area decreased by 0.8%, indicating superior prediction performance. The Shapley Additive Explanations (SHAP) value reveals that Hg, Sb, faults, As, and Au elements are closely associated with gold mineralization. Meanwhile, the catchment basin demarcates precisely the boundaries of elemental migration. This effectively minimizes false positive anomalies at the boundaries (formed during the gridding process) and makes the prediction results more reliable. As a result, five gold mineral prospectivity areas were delineated. | ||
These are possibly similar items as determined by title/reference text matching only.
