Abstract:
We live in a world were scientific and technical advances require increasingly specialised knowledge while drawing expertise from ever more diverse technical areas. In an effort to better understand the relationships between different areas of innovation, and the role of specialisation, diversity and ubiquity in national and regional economies, we have mined several million patent records from the European Patent Office, along with their classification codes, and used them to construct a network of “patent-space”. Patents provide a rich data set when studying innovation. Networks of scientific publications, such as that in [1], formed from inter-journal citations, illustrate the links between different disciplines as new knowledge is created, while networks of countries and the goods they export, such as the “product-space” network in [2], give insight into the economic complexity (or otherwise) and the likely areas of growth for national economies. Using patents allows us to take an intermediate view and investigate the role of science and innovation in economic growth. We take an approach similar to [2], identifying when individual countries or geographic regions have a “revealed comparative advantage” with respect to particular technical areas. We have constructed a proximity network as a base-map for the space of patentable innovation. We find that patent-space is heterogeneous and highly structured, and that the structure depends on the size or “granularity” of the regions that data is aggregated into. By overlaying data for particular regions on the patent-space base map we are able to explore temporal and regional trends – in particular how the innovation systems of different countries has produced quite different areas of specialisation. Figure 1: Patent-space for New Zealand (left) and South Korea (right) - two countries with very different innovation systems. Nodes represent patent classification classes. Nodes are dark when the country has a comparative advantage in that area and faded out otherwise. Nodes are connected if a comparative advantage with respect to one classification tends to occur in conjunction with a connected classification. [1] L. Leydesdorff & I. Rafols, Journal of the American Society for Information Science and Technology (2008). [2] C. Hidalgo & R. Hausmann, Proceedings of the National Academy of Sciences (2009).