Co-network Analysis Based On The Social Network Analysis Theory And Complex Network Theory

As the development of science and technology, the research gets more and more sophisticated.it is hard for researchers to complete a research or thesis alone. Scientific publications, as main output of scientific research, co-published by several authors are much more general, rather than single-authored publications. Collaboration-network which consists of the author with the edge as the Collaboration-relationship reflects the Collaboration between experts and scholars in scientific publications. Analysis of this network can reveal a lot of information about Collaboration structure between scholars, for example, who are co-authors of a scholar, whether a scholar in a core of this network, which coauthorships are stronger than others, What is the difference between their role in the network.From various perspectives, This information reflects scholars'research cooperation. It indicates that the network con tains a lot valuable information about scholars. Mining and identification of this information with in-depth analysis will play a significant role in guiding the formulation of science and technology management and technology policy.This thesis stems from discussion of the structure of the scientific collaboration network, utilizes t methods of complex network analysis and social network analysis, give out a analysis on the scientific collaboration network from both macro and micro perspectives, and in both dynamic and static way.This paper extracts16years of conference proceedings articles from1995to2011as an experimental data set, with the corresponding network known as collaboration network of data mining.Firstly, based on the quantitative graph theory to measure and analyze these qualitative changes in the cooperation network topology, such as density, diameter, and relative size of the largest component of the network.The analysis shows that the development of a field would experience structure topology evolution from relatively small disconnected components to large networks with huge connected component.Therefore, the number of edges and nodes in the largest component will experience evolution from relatively few to the many.In collaboration network, based on the relations and cooperation between coauthors, we evaluate importance of scholars in it, a.k.a.in what position scholars are, that is, the centrality of scholars.Based on analysis of the classic centrality metric, an improved node centrality metrics (c-index) is proposed, to measure collaboration strength of nodes in a weighted collaboration network.Concerning research direction distribution pattern, this article divides this question into three parts:a number of research direction, heterogeneity of the research direction of research uniformity of research direction.After proposing the concept of heterogeneity and uniformity of research direction, with the case of data mining disciplines, analysis focuses on the distribution pattern of the subject direction and classification of authors.We analysis collaboration network link prediction with classic algorithm based on node similarity measure and discuss the role of the weak link in the link prediction.In order to fully explore the scientific cooperation pattern, we introduce a time parameter, and define the concept of an affine group.As the result of the Analysis, some interesting trends in the of scientific cooperation are obtained, such as that the average size of the group grows exponentially, and the increasing of number of authors obeys the power-law, and by extrapolation it is able to determine the approximate date of a separate affine group. Also, spectral analysis-based approach can be used to divide communities in the network.These studies not only enriched the theory of collaboration network, also provides useful lessons for science and technology management, and science and technology policy formulation.