Abstract
Identifying emerging technologies is paramount for anticipating opportunities and risks in organizations early. Previous studies highlight that the combination of technologies sparks innovations. This phenomenon, coined technological convergence, is mainly studied using patent data. However, patents are registered at a late stage of technological development, which hinders early anticipation. Additionally, these studies do not quantify the degree of technological convergence. In this work, we address these two concerns by analyzing a type of document that arrives at an earlier stage and computing a technological convergence index. First, we extract a corpus of keywords from titles and abstracts for each arXiv computer science category subsection. Then, we create a dynamic proximity index between the subsections according to the number of keywords shared in their respective corpus. We study the cryptography and security subsections in which we observe three relevant processes: (i) a divergence with Information Theory, (ii) a stagnation with Databases, and (iii) a convergence with Machine Learning and Sound. Next, we apply transfer learning methods to forecast our index. Altogether, our work extends the TechMining literature by offering a novel method and informing decision-makers about early technological convergence.
Podcast
Research Project
Towards-a-Technology-Convergence-Index-for-InformationSource: VP Institute