Scholarlink

<p>Although knowledge graphs are powerful tools for representing heterogeneous geological knowledge in the mineral deposit domain, their potential for quantitative analysis of mineral deposit genesis for knowledge discovery remains underexplored. In this study, we developed a network-based methodology for knowledge discovery in large-scale knowledge graphs of mineral deposits. Taking gold deposits as a case study, we constructed a graph comprising over 4338 geological entities and 9547 semantic relationships, organized under a uniform schema. A community detection method based on modularity optimization was adopted to analyze the characteristics of the knowledge graph. At a resolution of 1.0, a modularity score of 0.59 was obtained. This process identified 21 communities, each of which can be interpreted in geological terms. The methodological contribution of this work lies in framing deposit classification as a graph partitioning problem, allowing genetic relationships among deposits to be inferred from graph-theoretic structures. Comparative analysis showed that this approach not only recovers traditional metallogenic types (e.g., magmatic–hydrothermal and Carlin-type deposits) but also uncovers new insights, such as supporting a shallow hydrothermal origin for the Zhilingtou gold deposit in Zhejiang. Our results demonstrate how network analysis of knowledge graphs can serve as a universal mathematical framework for classifying deposit genesis, validating metallogenic models, and guiding mineral exploration strategies. This study highlights the wider potential of graph-based community detection in geoscientific knowledge discovery.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Gold Deposit Knowledge Graph Mining Based on Network Analysis: Implications for the Gold Deposit Understanding and Classification

  • 期刊论文

  • 发表日期：

    2026年3月6日

查看全文
  • Bo Wang,
  • Mingguo Wang,
  • Chengbin Wang,
  • Jianguo Chen,
  • Liheng Chang,
  • Nan Li

摘要

Although knowledge graphs are powerful tools for representing heterogeneous geological knowledge in the mineral deposit domain, their potential for quantitative analysis of mineral deposit genesis for knowledge discovery remains underexplored. In this study, we developed a network-based methodology for knowledge discovery in large-scale knowledge graphs of mineral deposits. Taking gold deposits as a case study, we constructed a graph comprising over 4338 geological entities and 9547 semantic relationships, organized under a uniform schema. A community detection method based on modularity optimization was adopted to analyze the characteristics of the knowledge graph. At a resolution of 1.0, a modularity score of 0.59 was obtained. This process identified 21 communities, each of which can be interpreted in geological terms. The methodological contribution of this work lies in framing deposit classification as a graph partitioning problem, allowing genetic relationships among deposits to be inferred from graph-theoretic structures. Comparative analysis showed that this approach not only recovers traditional metallogenic types (e.g., magmatic–hydrothermal and Carlin-type deposits) but also uncovers new insights, such as supporting a shallow hydrothermal origin for the Zhilingtou gold deposit in Zhejiang. Our results demonstrate how network analysis of knowledge graphs can serve as a universal mathematical framework for classifying deposit genesis, validating metallogenic models, and guiding mineral exploration strategies. This study highlights the wider potential of graph-based community detection in geoscientific knowledge discovery.