Detecting Technology Portfolios in the Semiconductor Industry

Abstract

Semiconductor serves as the base for the Artificial Intelligence Society. We are interested in detecting technology trends and changes in the semiconductor industry because there are some large resource allocation and organization restructure in this industry recently. In this study, we narrowed our research boundary into the U.S. and Japan. We extracted patent information related with semiconductor and made a citation network. We used Louvain method to cluster the maximum connected component and considered several largest clusters. Results show that the technology portfolios among the two countries are different. We used “tf-idf” to detect keywords and features of these unbalanced clusters. In the future, we will link the applicant information in a patent database with mergers and acquisitions (M&A) information in a company database. We will compare and integrate findings from different sources, such as investment, M&A, technology features and industrial policies in order to have a comprehensive understanding.