Entity recognition model of power safety regulations knowledge graph based on BERT-BiLSTM-CRF

In the process of constructing the knowledge graph of power safety regulations, the traditional named entity recognition method is difficult to effectively identify the key information of the entity because the boundary of the power safety entity is fuzzy and difficult to define. Therefore, this paper proposes a power safety named entity recognition model based on BERT-BiLSTM-CRF. First, the word vector expression layer based on Transformer’s bidirectional encoder (BERT) obtains word-level features; then the bidirectional long-short-term memory neural network (BiLSTM) layer is used to extract contextual features to form a feature matrix, thereby improving the accuracy of text feature extraction; The optimal tag sequence is generated by the conditional random field layer (CRF), and the output result is corrected. Through the analysis of experimental examples, the validity and superiority of the proposed model are verified.