Purpose/Significance Guided by the key factors of clinical medication decision-making, to construct a drug knowledge graph, so as to provide accurate and structured knowledge support for rational drug use decision-making. Method/Process The information of the instructions for cardiovascular drugs is integrated, and fine-grained ontology modeling and multi-dimensional relationship attribute definition are carried out. A large language model-driven knowledge extraction and standardized alignment strategy is adopted to construct a drug knowledge graph for rational drug use, and an intelligent drug question-answering system is designed based on retrieval-augmented generation (RAG) technology. Result/Conclusion Taking cardiovascular drugs as examples, a drug knowledge graph containing 5 215 entities, 17 288 relationships and 9 598 relationship attributes is constructed, providing reliable data support and decision-making basis for rational drug use.