Research On The Relationship Deduction Method Of Knowledge Graph Of Aviation Safety Incidents

The knowledge graph organizes and processes real-world knowledge effectively in the form of graphs.The aviation safety incident management based on knowledge graph can effectively improve the aviation safety management level.At present,the research of general knowledge graph and domain knowledge graph mainly focuses on the construction of knowledge graph,but there are few high quality knowledge graph and application research.Therefore,this paper focuses on the knowledge graph relation deduction method and its application in aviation security field in the following aspects.On the basis of in-depth research on the knowledge graph relation deduction method,In view of the problems that the entity and relation are relatively independent in the presentation learning process of the Iter E network model,and the rich semantic information between entity and relation cannot be fully learned and the learning efficiency of rules is low.In this paper,a method of knowledge graph relation deduction is proposed to enhance the correlation between entity and relation by features,the model is called SIter E for short.Firstly,the Simpl E method is used to represent the entities and relationships in the knowledge graph.After that,the relationship features were superimposed in the way of add so that the features could be reused,effectively improving the generalization ability of the model.Then,the axiom set of knowledge graph is mined according to the axioms of OWL2,and the improved scoring mechanism is used to score the axioms,and the relation is deduced according to the high score axioms.The experimental results show that the SIter E method are better link prediction results have been achieved and higher rule learning efficiency on the public datasets,at the same time,this method can learn more high quality rules.Aiming at the problem of low quality of aviation safety incident knowledge graph constructed at present.Based on the historical events published in the accident database system of the Aviation Safety Network(ASN),analyzing the characteristics of datasets in the field of aviation safety and carrying out data processing and parameter selection.The SIter E method is applied to the relationship deduction of aviation safety incidents knowledge graph.The experimental results show that the SIter E method has a significant improvement over the Iter E method in the MRR,Hits@1,Hits@2 and Hits@3 indicators of the dataset in the aviation safety domain.In addition to digging out the potential relationships in the knowledge graph of aviation safety incidents,the quality of the knowledge graph of aviation safety incidents is improved continuously.