BERT-BiLSTM-CRF Chinese Resume Named Entity Recognition Combining Attention Mechanisms

Aiming at the traditional Chinese resume named entity extraction methods can not solve the problem of multiple meanings of a word well, as well as the problem of insufficient mining of potential semantic features of the context. In this paper, we propose a Chinese resume named entity recognition model based on the combination of Bidirectional Encoder Representations from Transformers (BERT), Bi-directional Long and Short Term Memory (BiLSTM) network and Conditional Random Field (CRF), and on the basis of which we introduce the Attention mechanism (Att). The input text is encoded at character level using the BERT pre-trained language model to obtain dynamic word vectors, and then the global semantic features are extracted using the Bi-directional Long Short Term Memory (BiLSTM) network, and then the Attention mechanism is used to assign the weights to better capture the key features, and finally the Conditional Random Fields (CRFs) are used to output the global optimal labeling sequences. The experimental results show that the BERT-BiLSTM-Att-CRF model proposed in this paper achieves better recognition results on the Chinese resume dataset.