Construction And Application Of Landslide Knowledge Graph

Landslide is a geological phenomenon endangering safety,which causes great losses to ecological environment,human life and property safety.In recent years,with the development of knowledge graph,it is of great significance to use knowledge graph in landslide field,effectively deal with multi-source landslide information,excavate landslide formation mechanism and spatial-temporal evolution process,provide high-precision and comprehensive knowledge management and decision support for disaster.However,the present knowledge representation model is not suitable to express the complex process and mechanism in landslide field.Data-driven spatial landslide prediction method is highly dependent on monitoring data,while knowledge-driven method is interfered by subjective factors.The geographic knowledge graph focuses on the construction,but the knowledge calculation based on knowledge graph has not been attempted yet.This paper explores the construction and application of landslide knowledge graph driven by data and knowledge,mainly from the following four aspects:Firstly,the landslide knowledge representation model considering spatial-temporal process and formation mechanism is studied.A hierarchical landslide knowledge representation model is constructed to reveal spatial-temporal process and formation mechanism with spatial-temporal framework and object as the core.Secondly,the techniques of knowledge extraction knowledge fusion and knowledge storage are used to process the relational data base of structured remote sensing image interpretation,disaster data of semi-structured website and unstructured landslide mechanism data,so the landslide knowledge graph is constructed.The query results of landslide knowledge graph were compared with the representation ability of mainstream knowledge graph.The results show that the landslide knowledge graph can effectively express the fact knowledge,process knowledge and mechanism knowledge.Thirdly,a landslide spatial prediction method driven by integrated data and knowledge is designed.Taking the loess landslide in Heifangtai area as the training sample,the probability of landslide in the area to be measured was judged.The precision analysis of landslide hidden danger is realized,which is suitable for small sample data structure and is conducive to improving the accuracy and applicability of prediction method.Finally,a landslide knowledge base question answering is designed based on the landslide knowledge graph stored by Neo4 j.The knowledge stored in the landslide knowledge graph is displayed in the form of intelligent question and answer.The knowledge base question answering can complete the intelligent question-answering of binary relation and multivariate relation,realize the query and reasoning function of knowledge graph,and provide practical application value for disaster knowledge management.