A Military Named Entity Recognition Method based on pre-training language model and BiLSTM-CRF

Abstract Military named entity recognition is the basis of the military intelligence analysis and operational information service. In order to solve the problems of inaccurate word segmentation, diverse forms and the lack of corpus in military texts, the author proposes a method of military named entity recognition based on Pre-training language model. On this basis, and taking advantage of Bi-directional Long Short-Term Memory (BiLSTM) neural network in dealing with the wide range of contextual information, the BERT-BiLSTM-CRF named entity recognition model was constructed. The experimental results on the tagged military text corpus show that the extraction effect of this method is better than that of the traditional methods.