그래프 신경망을 이용한 국방과학기술 융합 예측

Purpose This paper aims to predict the technology convergence in the defense field using link prediction models, including topological feature-based models and a graph neural network model. Methods Based on the patent data from 1979 to 2020, link prediction models for convergence technology have been presented. Topological feature-based models and a graph neural network model are implemented. Results A graph neural network model showed better performance than topological feature-based models, so future conversion networks are predicted based on the graph neural network model. As a result of the link prediction, battery technology will play an important role in connection with other military technologies in terms of technology convergence. Conclusion The main contribution of this paper is to show the future direction of defense R&D promotion from the viewpoint of technology convergence. As a defense technology strategically promoted by the government, investment is necessary to develop it into a mature technology. Therefore, the results can be used as a guideline when the government prioritizes investment strategies in the defense sector.