Abstract: This study proposes a method for identifying technological opportunities of high-value patents in colleges and universities, using theme modeling, mutation level method, machine learning and outlier detection algorithms to further identify technological themes and patented technologies with potential technological opportunities on the basis of evaluating high-value patents in colleges and universities. Taking the field of "Generative Artificial Intelligence" as an example for empirical evidence, the results show that the potential technology themes in the field of "Generative Artificial Intelligence" are centered on cutting-edge areas such as deep learning, neural networks and machine learning, and AI imaging and AI diagnosis and treatment are potential technological opportunities in this field, and the above technologies are vigorously supported by relevant national policies. This method can break through the core problems such as poor targeting of the identification results of a single technology opportunity identification method, low value of the identified patents, and a single form of the identification results, and the relevant identification results can provide decision-making support for the technology transfer, technology research and development, and technological innovation of universities.

Key words: High-value patents, Patent valuation, Technology opportunity identification, Catastrophe progression method, Local outlier factor