【摘要】 为了优化医疗领域的中文实体识别效果，提出一种基于BERT-BiGRU-CRF的理论模型。首先将原始的问答数据传入到BERT模型中，对原始的问答数据完成词向量表示，词向量表示的数值矩阵被传入到BiGRU模型进行获取语句的上下文特征，最后针对原BiGRU模型标签问题的错误判定使用CRF模型进行修改。通过试验对比，模型最终得到的F1值为76.39%，相比较其他模型有所提升。更多还原

【Abstract】 In order to optimize the effect of Chinese entity recognition in the medical field, a theoretical model based on BERTBiGRU-CRF is proposed. Firstly, the original Q&A data is passed into the BERT model, and the word vector representation is completed for the original Q&A data. Then, the numerical matrix represented by the word vector is passed into the BiGRU model to obtain the contextual features of the sentence. Finally, the CRF model is used to modify the error determination of the labeling problem in the original BiGRU model. Through experimental comparison, the final F1 value obtained by the model is 76.39%,which is improved compared with other models.更多还原