Named Entity Recognition for Chinese Aviation Security Incident Based on BiLSTM and CRF

Targeted at the issue of automatic acquisition of aviation security event entities. In this paper, the entity recognition model of aviation security incident based on character vector BiLSTM and CRF is constructed. In order to avoid word segmentation errors, character vectors are used in this paper. The sentence vector is obtained by PV -DM, and then the character vector and sentence vector are fused to make full use of the lexical information in the text. Then input BiLSTM to obtain the context characteristics of the text, and ensure the consistency of the output entity label through CRF. Finally, a data set of aviation security incidents is constructed, and the experimental results show that the proposed method improves the effectiveness of entity identification in the field of aviation security incidents.