The paper introduces the research status of traditional Chinese medicine (TCM) symptom and the application of knowledge graph technology in TCM field, and selects Chinese Terms in Traditional Chinese Medicine and Pharmacy as the source of symptom knowledge to build a knowledge base of TCM symptom. Descriptive statistics, association rule analysis and other data mining methods are used to elaborate the distribution rules of high-frequency syndromes and symptom information, as well as the association relationship between symptoms and symptom tacit knowledge. Neo4j is used to construct the TCM symptom knowledge graph and analyze the knowledge association relationship between syndromes, symptoms and attributes.