Abstract
Big data has brought new opportunities to landslide research; however, due to issues such as complex data types, diverse semantic relationships, and unclear sharing mechanisms, deep mining of landslide data remains limited, and the advantages of big data are difficult to realize in landslide research. This paper proposes a knowledge graph construction method for landslides oriented toward the engineering geology domain, which extracts, fuses, and structures multi-source heterogeneous landslide knowledge to enable querying, association, and reasoning of landslide knowledge big data. Using a combined top-down and bottom-up approach, landslide concepts and ontologies are divided, forming a landslide knowledge system based on ten major categories of knowledge: landslide field investigation, landslide assessment, landslide types, landslide geomorphological characteristics, landslide morphological characteristics, landslide disaster information, landslide activity status, landslide formation mechanisms, landslide stability analysis methods, and landslide prevention measures. A knowledge graph schema layer including a concept layer, attribute layer, relationship layer, rule layer, and instance layer is established; landslide knowledge information is extracted from broad data sources to establish a semantic network, and redundant knowledge is fused to construct the knowledge graph data layer; using the Neo4j platform to store landslide knowledge, knowledge visualization and retrieval are realized, providing new ideas and methods for landslide mechanism research and disaster prevention and mitigation. The proposed landslide knowledge graph construction method can be extended to research on knowledge graphs for other types of disasters, and can generate connections with other disciplinary fields, promoting deep interdisciplinary integration.
Full Text
Construction of a Landslide Knowledge Graph for Engineering Geology
XU Qiang, CUI Shenghua, HUANG Wei, PEI Xiangjun, FAN Xuanmei, AI Ying, ZHAO Weihua, LUO Yonghong, LUO Jing, LIU Ming, XIA Min, WANG Fei, PENG Dalei, ZHENG Guang, CHEN Wanlin
State Key Laboratory of Geohazard Prevention and Geoenvironment Protection, Chengdu University of Technology, Chengdu 610059
Abstract
Big data has created new opportunities for landslide research; however, due to challenges such as complex data types, diverse semantic relationships, and unclear sharing mechanisms, deep mining of landslide data remains limited, making it difficult to fully leverage the advantages of big data in landslide studies. This paper proposes a method for constructing a landslide knowledge graph for the field of engineering geology, which extracts, fuses, and structures multi-source heterogeneous landslide knowledge to enable querying, association, and reasoning of large-scale landslide knowledge data.
Using a combined top-down and bottom-up approach, landslide concepts and ontologies are classified to form a landslide knowledge system based on ten major categories: landslide field investigation, landslide assessment, landslide types, landslide geomorphological characteristics, landslide morphological features, landslide hazard information, landslide activity status, landslide causal mechanisms, landslide stability analysis methods, and landslide prevention measures. A knowledge graph schema layer is established, comprising a concept layer, property layer, relationship layer, rule layer, and instance layer. Landslide knowledge information is extracted from extensive data sources to build a semantic network, and redundant knowledge is fused to construct the data layer of the knowledge graph.
The Neo4j platform is utilized to store landslide knowledge, enabling knowledge visualization and retrieval, which provides new ideas and methods for landslide mechanism research and disaster prevention and mitigation. The proposed method for constructing landslide knowledge graphs can be extended to research on knowledge graphs for other types of disasters and can connect with other disciplinary fields, thereby promoting deep interdisciplinary integration.
Keywords: Knowledge Graph; Landslide; Neo4j Graph Database