Named Entity Recognition Method based on Bert-Bilstm-Attention-CRF for Education field

With the rapid development of the new generation of information technology, the educational information system has accumulated a large amount of educational data information, which makes learners have low cognitive efficiency, difficulty concentrating, and eventually deviate from learning goals and fail to complete specific learning tasks. This paper proposes a named entity recognition method in the education field combined with Bert-BiLSTM-Attention-CRE to improve learners' learning efficiency. Firstly, BERT language model is used to extract text features to obtain the word granularity vector matrix; Then use Bi-directional Long Short-Term Memory (BiLSTM) to extract the word-to-word relationship between the input statements and the context; Thirdly, Attention mechanism is introduced to better utilize local contextual clues by assigning different weights to the text feature vectors, and strengthen the correlation between current information and context information; Finally uses the Conditional Random Field (CRF) model to extract the globally optimal output label sequence according to the dependency between labels; and the named entity of the education field is obtained in the end. The experimental results show that the Precision, Recall and F1 value of the model are 82.25%, 79.73%, 80.39% respectively, the overall recognition performance is high, and the effect is better than that of BiLSTM-CRF, CNN-BiLSTMCRF, HIDCNN, BERT-BiLSTM-CRF.