Building of Gas Emergency Knowledge Graph Based on Joint Extraction Method

In recent years, gas accidents occur frequently. In order to reduce the casualties and property losses brought about by gas disasters, emergency disposal needs to be scientific and rapid. Currently, gas accident cases are usually stored in the form of documents. The relation between case text and disposal measures is poor. So, decision-making in emergencies is difficult to meet the need of scientific and rapid. In this paper, a top-down gas emergency knowledge graph construction mode is used, and an extracting method of entity-relationship triples using the E-CasRel joint extraction model is proposed to construct the gas emergency knowledge graph for gas accident cases. The feasibility of the proposed method is verified by real cases. This study could enable the disposers to make scientific decisions in a short time and provide a meaningful method to guide the emergency disposal of gas accidents in future.