Abstract:Potato is the world"s fourth largest staple crop with high yield potential. To meet the future challenges of food security, selection of early-maturing and high-yielding potato varieties with stable disease resistance is needed. In this study, in order to assist the research of new potato variety selection and breeding, and to clarify the current status of new potato varieties in China, 227 potato selection and breeding literatures in the China Knowledge Network (CNKI) database were used as research objects, and a knowledge map was constructed by using GraphRAG and Qwen2-70B-instruct and visualized by using Gephi. Based on the constructed knowledge graph, the genealogy, resistance and fertility period of new potato varieties selected and bred in China in recent years were analyzed, and the analysis results indicated that more parents used in the selection and breeding of new potato varieties in recent years were Jizhangyan No. 8, Svante, Longyan No. 7, Feiuretta, and Early Large White, etc. Most of the new potato varieties were resistant to late blight, and most of them were of medium-late to late-maturity and late-ripening fertility. In summary, this study explored the realization path of using a large language model to rapidly construct a knowledge graph for potato new variety selection and breeding research, and analyzed 227 potatoes to provide a reference for the future discovery and utilization of potato germplasm resources.