Classification of Traditional Chinese Medicine Cases based on Character-level Bert and Deep Learning

As one of the traditional cultures of our country, Traditional Chinese Medicine (TCM) has received more and more attention. As a valuable asset inherited from ancient times, TCM medical cases carry the core knowledge content of TCM. Accurate medical case classification is an important part of establishing a correct medical case diagnosis and treatment system, and is also an important part of medical assistance system. This paper proposes a new model to effectively classify medical cases. First, the multi-layer semantic expansion method is used to increase the semantic information of TCM medical cases in instance layer and attribute layer. Then, the character-level Bidirectional Encoder Representations from Transformers (Bert) model is used as a language model for text representation of the medical cases, and the results are as the input of the deep learning models. Finally, the optimized Text-Convolutional Neural Network (Text-CNN) model is used to classify TCM medical cases, and the reliability and accuracy of the whole model are verified through the comparison with the result of other text representation and classification methods.