Global Multi-Level Analysis of the 'Scientific Food Web'
We introduce a network-based index analyzing excess scientific production and consumption to perform a comprehensive global analysis of scholarly knowledge production and diffusion on the level of continents, countries, and cities. Compared to measures of scientific production and consumption such as number of publications or citation rates, our network-based citation analysis offers a more differentiated picture of the 'ecosystem of science'. Quantifying knowledge flows between 2000 and 2009,
... e identify global sources and sinks of knowledge production. Our knowledge flow index reveals, where ideas are born and consumed, thereby defining a global 'scientific food web'. While Asia is quickly catching up in terms of publications and citation rates, we find that its dependence on knowledge consumption has further increased. P aper and citation counts are the 'official currency' in science and are widely used to assess the productivity and impact of authors, institutions, and scientific fields 1-5 . Many academic rankings focus on numbers P(t) of publications in leading journals and citations rates C(t), i.e., on knowledge production and consumption over time t. Examples are rankings of people, institutions, cities, or journals 6-9 . They show that new powers such as China and Brazil have recently emerged on the global scientific landscape 10 . Extrapolating these trends, it seems that the USA and Europe might lose their academic leadership. However, academic leadership requires one to be first to publish a paper and others to cite the ideas. Simple counts of publications and citations of an entity (be it an author, institution, city, geographic area, journal, or scientific field) do not reveal who cites whom (thereby consuming knowledge from others), and who is cited (i.e., who produces knowledge consumed by others). The network-based approach proposed here assumes the existence of a 'scientific food web' that interconnects academic entities via knowledge flows. A network perspective is important, because in many complex systems (such as the scientific ecosystem), interaction effects can be more relevant for the resulting system behavior than the properties of the interacting entities themselves. For example, it has been shown that author teams manage to be more successful than single authors 11-14 . The social, network-based character of knowledge diffusion underlines this perspective as well    . Compared with other ecosystems 18, 19 , an entity in the scientific food web is considered to be particularly successful ('fit'), if its knowledge is consumed (cited) more than expected. The analogy to ecosystems is chosen here to pronounce the mutual interdependencies and synergy effects in knowledge creation, since the production of new knowledge is nourished by the previous existence of relevant knowledge sets and their recombination. This is in line with research that uses the concept of ecosystems to shed new light on financial markets 20 and the evolution of national economies 21 . In previous work, networks of scientific papers 22 were used to analyze the evolution of scientific fields 23 , to study innovation diffusion 24,25 or clickstream patterns 26 , and to model the emergence and development of scientific fields 27 . Moreover, knowledge diffusion has been mapped between 500 major U.S. academic institutions, using a 20-year dataset of 47, 073 PNAS papers 28 . Other research studied knowledge import patterns for the field of transportation 29 . Our current study goes significantly beyond this by proposing and validating a new networkbased index measuring higher-than-expected knowledge flows, which can be consistently applied on multiple levels. We demonstrate this by evaluating 13 million papers to identify global trends of knowledge diffusion at the level of continents, countries, and cities.