Research on Disruptive Patent Identification from the Perspective of Technology Convergence
[Purpose/Significance]In the context of the technological revolution and industrial transformation,disruptive technology is a new way to easily lead to technological breakthroughs and subvert traditional game rules.It is of great strategic significance to actively identify and layout disruptive technologies to realize leapfrog devel-opment in national scientific and technological innovation.[Method/Process]It proposed a disruptive technology prediction method from the perspective of technology convergence.Firstly,based on the co-occurrence relationship of patent IPC,it constructed a technology domain fusion network through a time series approach,and identified the disruptive domain convergence pairs in the domain fusion network with structure entropy indicators.Secondly,it mapped convergence disruptive fields to patents to obtain candidate disruptive patents.On this basis,through in-depth analysis of the core characteristics of disruptive patents,it explored effective measurement indicators.Finally,based on the deep learning Tabnet model,it constructed the correlation between patent indicators and disruptive patents to achieve disruptive technology prediction.[Result/Conclusion]Taking the field of artificial intelligence as an example,it verifies the feasibility and effectiveness of the method.The research results obtain a total of 1025 disruptive patents.Among them,the disruptive convergence field(G06K9,G06N3)contains 443 disruptive patents,involving in four major disruptive directions:smart healthcare,smart transportation,smart security,and smart man-ufacturing.It can provide reference for the technological layout and strategic formulation of stakeholders such as the government,industry,and enterprises.
technology convergencedisruptive patent identificationstructure entropydeep learningTabnet model
万方数据
维普
