Purpose/Significance To construct the knowledge graph of Chinese patent medicine instructions, and to realize intelligent recommendation based on knowledge reasoning. Method/Process Based on the technology of Py2neo combined with Neo4j graph database to build knowledge graph, the information of 1 380 kinds of Chinese patent medicine are sorted out and the database is built. Knowledge reasoning is realized based on the triadic closure algorithm, and the recommendation score of Chinese patent medicine is calculated by the probability model. Result/Conclusion In the study, 11 103 entity concepts and 24 semantic relationships are formed. The knowledge graph of intelligent recommendation of Chinese patent medicine is constructed, and the intelligent recommendation platform of Chinese patent medicine is built to realize accurate recommendation of Chinese patent medicine. The combination of Chinese patent medicine and knowledge graph is realized, which provides a method reference for traditional Chinese medicine (TCM) auxiliary diagnosis and treatment, and provides references for further research on knowledge visualization of Chinese patent medicine.