Network reconstruction is a hot problem in current research, which is the process of mining the network topology. In this study, we present a new approach for reconstructing trade dependence networks using a panel regression model, with a focus on semiconductor products as a case study. We establish the econometric model under the framework of dynamic network reconstruction and panel regression. Parameters are estimated using the Adaptive Signal Lasso method, and we propose a two-step procedure for testing the significance of variables, utilizing trade data spanning from 2001 to 2020. Based on the parameter estimation results and complex network analysis, we construct the trade-dependent network and study its topological characteristics in detail. This study enriches the methods of economic network reconstruction, and provides new research ideas for the analysis of economic problems. At the same time, our findings offer policy suggestions for the economic and trade development of China’s high-tech products.