Technology-Transferability Analysis of Universities and Public Research Institutes Using Deep Neural Networks

Abstract

Both academia and business are paying close attention to the commercialization of technologies patented by universities and public research institutes (UPRIs). Previous research has mostly focused on the organizational characteristics of technology transfer offices (TTOs) that influence technology transfer, as TTOs are the primary stakeholders in the commercialization process; however, most past studies have rarely focused on the actual UPRI technology, the object of the intended transfer. Therefore, this study proposes a UPRI technology-transferability prediction model for determining the linkages between UPRI technologies and technology transfer. The steps used in this study are as follows: 1) identifying transferred and non-transferred UPRI technologies using the global research identifier database and the United States patent assignment data; 2) defining patent indicators related to UPRI technology characteristics; 3) building a UPRI technology-transferability prediction model based on the application of deep learning to patent right transfer data; and 4) validating the constructed model’s performance and comparing it with baseline models. We then discuss the differences between UPRI and private firm technologies, as well as the technological factors that affect the transferability of UPRI technologies. The proposed model and findings can be used by UPRIs to better understand and commercialize their technologies.

Keywords

• Deep neural networks
• patent analysis
• patent transferability
• university and public research institute