Analysis Of Scientific Collaboration Network Based On "Chinese Science Bulletin"

With the development of complex network theory, empirical research based on it has been always making progresses. At the same time, scientific collaboration networks (SCN) which have drawn more and more attentions among scholars are widely analyzed. Many scholars have done a series of researches on SCN of other countries in the world. They tried to excavate the collaboration patterns of scientists and detect their behavior features. This thesis aims for detecting the collaboration relationship among Chinese scientists. We constructed a collaboration network based on the papers published on Chinese Science Bulletin during the past 20 years (1988-2007). Static properties have been concretely analyzed and the evolving process is also observed. Basing on these characters, we have proposed an evolving model, explains the forming mechanism. By comparing with the empirical results, the rationality of the model is expounded and forming mechanism of the real-life network is further explained.The main contributions of the dissertation are summarized as follows.1 We have summarized the main work about SCN from two aspects: empirical analysis and model research. This thesis illustrates the construction way of SCN, introduces the methods and results of some important empirical researches and investigates several models of SCN.2 We have collected the collaboration information of authors who have published papers on Chinese Scientific Bulletin in recent 20(1988-2007) years. We have analyzed the distribution of authors'productivities and evolving feature of the collaborating sizes. We have constructed a collaboration network and observed its static and dynamic properties. We have found the network possesses notable clustering, small-world and assortative mixing properties and community and hierarchical structures. At last we have discovered that the strength distribution is more close to power-law than the degree distribution and authors repeat collaborations.3 According to the results of the empirical analysis, we have proposed a Scientific Collaboration Evolving Network Model, introducing two kind of connecting mechanism just like BA model: growing and preferential attachment. By changing the global preferential attachment probability we can modulate the strength preferential mechanism and collaboration experience preferential mechanism. We have studied how the network'static and dynamic properties change with variable parameter of evolving the model. We have got a conclusion that the strength preferential mechanism plays a main role in the network evolution and the experience preferential mechanism also affects the process.