Research On Key Technologies Of Knowledge Graph Of Urban Rainstorm Waterlogging Based On Public Opinion

Urban rainstorm waterlogging is one of the most common natural disasters in China,which poses a huge threat to the personal safety and property of the people.Traditional methods of collecting disaster information are difficult to obtain real-time information from the disaster area in a short period of time,while a large amount of public opinion information from the front line of the disaster area emerged on the Internet in the first time,which can provide timely information for emergency disaster response and decision-making,and has important significance for emergency disaster rescue.Based on Internet public opinion information,combined with knowledge in the field of geology and natural disasters,this thesis introduces the theory of granular computing to construct a multi-granularity knowledge graph of urban rainstorm flooding,studies the knowledge inference method that takes into account semantic and structural information.This paper combines knowledge graph technology with the Internet public opinion information,and timely and effectively obtains disaster information,providing technical support for improving the efficiency of disaster response and decision-making of urban rainstorm waterlogging.The main contents of thesis include the following aspects:(1)A method of constructing multi-granularity knowledge graph of urban rainstorm waterlogging.In response to the need for rapid construction of domain knowledge graphs of urban rainstorm waterlogging disaster,this paper builds on the characteristics of urban rainstorm flooding and based on natural disaster system theory and simple event models to establish six elements and five categories of attribute representation for urban rainstorm flooding events.and designs the ontology repository of urban rainstorm flooding knowledge from top to bottom.The granular computing theory is introduced to establish a hierarchical recursive quotient space model for the knowledge graph,and a multi-granularity knowledge graph of urban rainstorm waterlogging based on ontology is constructed,laying the foundation for knowledge graph inference.(2)A knowledge graph inference method that takes into account semantic and structural information.In response to the timely warning needs of urban rainstorm flooding disasters,and in view of the existing problem that the knowledge inference method of the Relational Graph Convolutional Network(R-GCN)only collects the node features of the knowledge graph,without collecting and utilizing the knowledge graph structure information and entity description information,this paper proposes a knowledge graph inference method that takes into account semantic and structural information.It utilizes SR-GCN+GSA(Semantic Relational-Graph Convolutional Network+ Graph SAmple and aggregate)model to learn node features of neighboring entities,semantic information of entity descriptions and structural information of the knowledge graph,which can improve the accuracy of knowledge graph inference.(3)Case analysis of urban rainstorm waterlogging disaster warning.Based on the research of this paper on the construction method and knowledge graph inference method of urban rainstorm flooding knowledge graph based on public opinion,taking the "Zhengzhou ’7.20 Super Rainstorm’ Urban Waterlogging" as a specific example,through experiments and analysis,the method of constructing knowledge graphs and the knowledge inference model of taking semantic and structural information into account are validated in the flooding disasters of Urban rainstorm waterlogging.