Empirical analysis of the coauthorship network based on DBLP

To identify and analyze the collaboration pattern between scientists in computer science discipline, we constructed a new coauthorship network by extracting the article information of the whole year of 2012 from DBLP library. The nodes are scientists and two scientists are connected if they have coauthored a paper. We study the network in depth by the theory of complex network. The empirical analysis shows that the productivity of authors follows a power-law distribution and the size of collaboration follows an approximate exponential distribution. Moreover, we find that the network is a small-world network and have no apparent scale-free property. Finally, four different indicators are employed to study the impact of authors.