[Purpose/Significance]The study focuses on the development of library knowledge mining services in the era of large language models. It explores the integration of traditional retrieval methods with retrieval augmented thoughts（RAT） to provide users with both profound and logically rigorous knowledge information, aiming to enhance the quality of library knowledge services.[Method/Process]The study adopted an experimental research approach, taking the Name Authority Database of Shanghai Library as the data source. It specifically focused on mining information related to tea culture in Song Dynasty, uncovering more details about this historical period from library collections, thus offering users a more comprehensive and in-depth perspective.[Result/Conclusion]Experimental results indicate that the RAT method has certain advantages in library knowledge mining services. This approach not only improves the efficiency and accuracy of data mining but also provides new technical solutions for library knowledge mining, offering valuable insights for the development of smart libraries.