OBJECTIVE: Chronic kidney disease has become one of the major public health problems in the world. Traditional Chinese medicine ancient literature have a lot of records related to chronic kidney disease. Machine learning data mining provides a new method for ancient literature mining, however there are many things need to be improved in ancient knowledge mining.METHODS: This study explored the ancient books of renal edema by knowledge graph, and deeply related the most important medicine, from the summary of core drugs - drug-disease knowledge graph and network construction - drug association cluster analysis —recessive knowledge discovery” has compiled the experience of ancient and modern doctors in the treatment of renal edema. Provide new ideas for further research and clinical practice.RESULTS: The drug distribution graph showed that Poria cocos was the most frequently used herbal medicine, and its clinical practice was the most closely related to the symptoms of “swelling below the waist”, “asthma”, “difficult in speaking”, “Jian” and “Zhengsong”. In addition to the known effects of water soaking, spleen and soothe the nerves, the related symptoms of sputum are further explored. It is also used as a drug in clinical excavation in the excavation of ancient books and the experience of famous Chinese medicine practitioners, in order to achieve the purpose of eliminating wind evil.Conclusion: The ontology knowledge base and auxiliary mining system of traditional Chinese medicine ancient books based on semantic network have unique advantages for discovering tacit knowledge and broadening the clinical understanding of Poria cocos. It can further explore the tacit knowledge of ancient literature through the knowledge graph, and visualize the explicit knowledge of the ancient doctors" dialectical use and the experience of drug use, which will help the full exploitation and utilization of ancient books resources.