텍스트마이닝 기반 학계와 산업계의 기술보호 연구 동향 비교: 주제 변화 양상과 시차 구조 분석

This study employs text mining techniques to examine academic and industrial trends in the field of technology protection and to identify the evolution of thematic structures and temporal lag relationships between the two domains. An analytical framework integrating TF-IDF-based keyword extraction, SBERT embeddings, and K-means clustering was developed to enable semantic topic identification. Annual variations in cluster proportions were analyzed to capture topic evolution, and Shannon entropy was used to assess topical diversity. Cross-correlation analysis was further conducted to verify topic-specific lagged diffusion patterns between academia and industry. Using academic publications from OpenAlex and industrial patents from the EPO DOCDB over the past decade, the analysis reveals that academic research exhibits a dispersed structure centered on cybersecurity, design frameworks, and emerging technologies. In contrast, industrial activity shows a concentrated focus on practical domains such as authentication and access control, threat response, semiconductors and materials, and AI-blockchain security. While several topics show that academic research precedes industrial activity by one to two years, areas such as technological innovation and network or cloud security exhibit simultaneous responses across the two sectors. Overall, this study provides an integrated depiction of the structural evolution of technology protection and the knowledge transfer mechanisms linking academia and industry. The findings offer empirical evidence that can support differentiated technology protection policies and R&D strategies aligned with domain-specific characteristics.