A novel medical case entity recognition method based on the BiSRU-TextCNN

the traditional BiLSTM neural network model can't extract local semantic features. What’s more, the model training focuses on relevant data and ignores irrelevant data as much as possible. Therefore, the novel medical case entity recognition called method called BiSRU-TextCNN solves the problem. This method fuses multivariate feature vectors, fully extracts local features and global features of related data. The experimental results show that the precision, recall rate and F1-Measure value of the new model are significantly improved, and the training time of the model is obviously shortened.