A BERT-BiLSTM-CRF Model for Chinese Electronic Medical Records Named Entity Recognition

Named entity recognition is a fundamental task in natural language processing and many studies have done about it in recent decades. Previous word representation methods represent words as a single vector of multiple dimensions, which ignore the ambiguity of the character in Chinese. To solve this problem, we apply a BERT-BiLSTM-CRF model to Chinese electronic medical records named entity recognition in this paper. This model enhances the semantic representation of words by using BERT pre-trained language model, then we combine a BiLSTM network with CRF layer, and the word vector is used as the input for training. To evaluate the performance, we compare this model with several baseline models in CCKS 2017 datasets. Experimental results demonstrate that the BERT-BiLSTM-CRF model could achieve a better performance than the other baseline models.