Potato ranks as the world's fourth-largest staple crop exhibiting high yield potential. Selection of early-maturing， high-yielding potato varieties with stable disease resistance is an urgent need to meet future food security challenges. In this study， to support new cultivar development and elucidate current potato breeding trends in China， 227 potato breeding literatures from the China National Knowledge Infrastructure （CNKI） database were used. We deployed GraphRAG and Qwen2-70B-instruct to construct a knowledge graph， which was subsequently visualized by using Gephi. Based on the constructed knowledge graph， we analyzed the genealogy， resistance and fertility period of new potato varieties selected and bred in China from 2004 to 2024. Our findings revealed that prominent parental lines in breeding programs included Ji Zhang Yam 8， Svante， Feiuretta， and Early Large White， etc. Most potato cultivars were resistant to late blight， with the majority classified as mid-late to late maturity types. 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， providing a reference for the future discovery and utilization of potato germplasm resources.