Integrating Persistence Process into the Analysis of Technology Convergence Formation Using STERGM

Abstract

Understanding the dynamics of technological convergence is indispensable for both academic and industrial perspectives. Traditional analyses have mainly focused on the formation process, often overlooking the role that persistence process plays in shaping technology networks. This paper endeavors to fill this gap by incorporating the persistence process into the analysis of technological convergence using the Separate Temporal Exponential Random Graph Model (STERGM). This study evaluates both formation and persistence processes using a decade-long dataset of patents related to breast cancer drugs, providing a more comprehensive view of technological convergence. Two core research questions are addressed, focusing on the dynamic mechanisms underlying technological convergence and its predictive capabilities. A significant finding of this study is that the network effects in formation and persistence processes differ, suggesting that focusing solely on one may not accurately capture the evolution of technology networks nor effectively grasp the process of technological convergence. In empirical forecasting, the model encompassing both formation and persistence processes yielded an F1 score of 68.34% for predictions, affirming the model's practical utility. Particularly noteworthy is that 'premature predictions' were observed in prediction errors, indicating that these instances may represent future potential converging technologies. In summary, this paper advances the analysis of technological convergence by considering both formation and persistence processes, and demonstrates that such a comprehensive approach significantly improves predictive performance.

Keywords: Technology Convergence, STERGM, Formation Process, Persistence Process, Predictive Performance