특허 피인용 네트워크 분석을 통한 혁신확산 패턴 분석: 가트너하이프사이클 AI 분야를 중심으로

In the midst of intensifying technological competition between nations and enterprises alike, the competition to maintain the technological lead persists as an ongoing trend. In light of this, both governmental bodies and industrial sectors are endeavoring to facilitate technology strategies and projections that are grounded in a substantial evidential basis. The “Hype Cycle”, which Gartner annually releases, is appraised as a pragmatic report pertaining to future technologies, yet the empirical examination of the Hype Cycle remains notably insufficient. The present research analytically investigates the differences in the characteristics of technologies at each stage of the Hype Cycle, grounded in the theory of innovation diffusion. In this study, based on innovation diffusion theory, we analyzed the differences in characteristics of 29 technologies in the artificial intelligence field at each stage of their hype cycle. We collected 4,018 registered patents from Korea, the United States, Japan, China, and Europe. In order to confirm the process and pattern of patent diffusion by stage, we attempted to analyze the relationship between patents that influenced subsequent patents by converting the citation information of applicants and examiners for each patent into network data. Based on this, we identified the scale of agglomeration and examined the structural characteristics of the network at each stage by dividing the degree of connection centralization and proximity centralization into outward and inward. In addition, the location of the network at each stage within the entire network was visualized to confirm differences in locational characteristics at each stage. Findings from the research demonstrate that the star-shaped network structure-characterized by inbound link centrality and closeness centrality values-intensifies, particularly in the ‘Tb(Innovation Trigger)’ and ‘S(Slope of Enlightment)’ stages. Especially at the S stage, this form was more robust, which can be interpreted as being influenced by the technological/business maturity, an inherent component of the Hype Cycle curve, derived from the patent data characteristics.