Building biobligraphic collaborating networks in an academic ranking system using the network deconvolution method

Abstract

To complete profound and large-scale research projects, scientists collaborate with each other. Scientist may cooperate in two ways, one is that some scientists directly work on a certain project and publish their findings jointly, the other is that scientists base their studies on predecessors’ and acknowledge that in the citations. In addition to abstractly visualizing the two forms of cooperation among researchers and quantifying the academic reputation of scientists in terms of publications, constructing a researcher cooperating networks could reveal the potential links among the subjects. This project proposes a viable approach to connect the scientists by their coauthoring publications, and a weighting method. It enhances the weighting criteria proposed by Newman (2001), then down-weights the edges to infer the direct connections with a cutting-edge method. The approach and process could serve as a framework for building the citation networks in the future. Based on the Computer Science Academic Ranking system, this project retrieves information from 36344 publications in class A and B conferences in the field of computer science to build a cooperating network for the authors, and is proved to be effective and efficient.