The structure, exchange, and transfer of knowledge in socio-technical systems

Abstract

This thesis aims to improve our understanding of the role of knowledge in economics and science. We analyze collaboration activities in these two domains, and show how the interactions among firms and among scientists influence the structure and the exchange of knowledge. We also model how the knowledge of these actors defines their collaborations. We show that knowledge is not only a consequence, but also a determinant of collaborations. To capture this interplay, we combine a statistical analysis of patent and publication data with agent-based models of collaboration activities. We follow a data-driven approach to study the structure, exchange, and transfer of knowledge. Specifically, using publication data we proxy the structure of scientific knowledge by reconstructing the citation network between publications. On this network, we quantitatively show that citation patterns strongly differ across time and scientific fields. We also identify the different knowledge of scientists, and quantify their knowledge exchange occurring during collaborations. Similarly, we use patent data to identify firms' knowledge and the knowledge exchange between firms involved in R\&D alliances. Then, to study the transfer of knowledge, we re-construct scientists' career paths by tracing their affiliations reported on their publications. With these paths, we construct the global migration network of scientists at city level, and analyze its topological properties. After analyzing collaborations activities, the exchange, and the transfer of knowledge, we reproduce these using agent-based models that we calibrate an validate against real-world data. In order to capture the very different processes behind these phenomena, we develop three different models. Precisely, to model collaborations activities among firms and their subsequent knowledge exchange, we combine and extend two existing models that captured only one of these phenomena each. Our a new model, instead, is able to simultaneously reproduce both these phenomena. To show how the knowledge differences between scientists determine their collaboration activities, we develop a second model that takes as input only these differences. Then, to model the transfer of knowledge, we develop a third agent-based model that reproduces scientists' migration at city level and the observed topological properties of the global migration network. Finally, we show that citation patterns between journals and scientists' career paths are better modeled by a new mathematical framework defined by higher-order networks than by traditional network models. By this, we challenge the application of the traditional network perspective to model the flow of knowledge between journals and the transfer of knowledge across research institutes.