Abstract:The breakthrough of "bottleneck" challenges is a crucial step towards achieving technological self-reliance, building a high-quality scientific and technological innovation system. Identifying "Bottleneck" key technologies provides practical evidence for breakthrough paths and organizational modes. Firstly, based on the general technological characteristics of key technologies and the strategic demands of "bottleneck" key technologies, a "bottleneck" key technology identification index system is constructed, consisting of 12 subdivided indicators in two dimensions. Then, natural language processing techniques and machine learning are employed for data feature mining, creating a method for identifying "bottleneck" key technologies. Train a machine learning classifier which is used to identify 305 "bottleneck" key technologies in the field of analytical instruments. Through LDA topic clustering analysis, 4 key technological areas of "bottleneck" challenges are identified. The results reveal that the diffusion of "bottleneck" key technologies in the field of analytical instruments is limited to technology alliances led by the United States, and China"s technological layout is extensive but unevenly developed. Plasma processing equipment, mass spectrometers, and nuclear magnetic resonance spectrometers are identified as key "bottleneck" areas in analytical instruments. Finally, based on the research findings, recommendations are proposed for the development of analytical instruments in China.