[Purpose/significance] Identifying and reasonably applying dark data is beneficial for maximizing data val⁃ ue. [Method/process] Taking user retrieval behavior as an example, based on the elaboration of problem reconstruc⁃ tion ideas and methods, a framework design is carried out for specific application scenarios in public libraries. Key breakthrough technologies are discussed from three aspects: retrieval behavior reconstruction, dark data recognition, dark data evaluation and grading. [Result/conclusion] Based on the above ideas and methods, empirical research was conducted to identify and process dark data in the Wenjin search system of the National Library. The feasibility of rec⁃ ognizing and applying dark data in public libraries was demonstrated by constructing a user retrieval intention data⁃ base, constructing a user retrieval trend analysis tool, and implementing data protection through reverse utilization of dark data.