Natural language processing methods are widely used to study the relationship between traditional Chinese medicine (TCM) prescriptions and diseases in textual data, and the results can discover the essence of TCM literature. In this paper, we get TCM treatment information from the abstract text at first by using the web crawlers. Second, the eigenvectors will be selected from the cleaned abstract text through syntactic analysis and feature extraction method of natural language processing. Then, artificial labeling part of the data is carried out based on different SVM classification models. The SVM classifier with TF-IDF feature vector is better. Last, we use the trained classifiers to classify all the data to construct the relationship between TCM prescriptions and diseases by performing neural network training, and predict the prescription for the treatment of the disease with the trained model. The result shows that our method has a certain positive effect on the research of Chinese medicine treatment diseases.