Construction of Rail Transit Network Fault Knowledge Graph Based on Pseudo-Dynamic Relationship Ontology Architecture

In this paper, an approach to construction of rail transit network fault knowledge graph based on pseudo-dynamic relationship ontology architecture is proposed. Firstly, an ontology architecture that includes pseudo-dynamic relationships is employed to completely separate fault patterns from fault entities without losing semantics. Secondly, a named entity recognition method based on the BERT-BiGRU-CRF model is adopted, which performs well in short entity extraction tasks. Thirdly, a pseudo-dynamic relationship extraction method based on the BERT-MEA(multi entities attention) model is adopted to extract static simple relationships, followed by treating triples as head and tail entities to determine pseudo-dynamic relationships. The implementation of rail transit network fault knowledge graph demonstrates the effectiveness of the proposed approach.