Research On Node Importance Measurement Of Large Software Network

With the development and progress of science and technology,more and more social networks show the characteristics of complex networks.As a new field of scientific research,the empirical research on complex networks has been paid much attention by many experts and scholars.On the one hand,the development of complex network provides great convenience for people's lives,but on the other hand,the complex system of the network to the people's life has brought some negative effects,such as large area power outages,network attacks,traffic gridlock,rumors spread,the spread of disease and so on.Accordingly,the identification of important nodes in complex networks plays a vital role in the operation of the whole complex network system in the study of complex networks.Although there are many ways to identify important nodes in complex networks,different identification algorithms are more or less limited by different indicators and different network types.In this paper,based on the past important node recognition algorithm,we take the large software network as the research goal,and get the important node recognition algorithm suitable for large-scale software network.The main research work of this paper is as follows:1.Some classical algorithms of complex network node importance are introduced,including Degree centrality,Betweenness centrality and Closeness centrality,K-shell decomposition method,Eigenvector centrality,PageRank algorithm,and analyzing the advantages and disadvantages of various algorithms.2.By analyzing the structure of complex software network,two kinds of new complex network static characteristic quantities-duplicato-degree and neighbor clustering coefficient.are proposed.The distribution of duplicato-degree and neighbor clustering of 10 large-scale open source software networks is analyzed.The results show that the duplicato-degree and neighbor clustering coefficient can better indicate the static characteristics of complex networks.3.On the basis of traditional sorting algorithm,a node importance metric p(i)based on local and global characteristics of network topology is proposed.It is based on the local information of the node itself and its neighbor nodes,and balances the local property of the measurement method by the degree centrality of the node.Then we selected four large open source software network to verify the measurement method,experimental results show that the importance of the new node node importance metrics p(i)for large software network assessment has high validity,the large-scale application software network shows outstanding measure effect.4.A new sort of node importance ranking indexp_i~*is established by using the duplicato-degree and the neighbor clustering coefficient.There are 10 kinds of large software networks that are selected to verify the effect of its node sorting on the software network.At the same time,the effect of node importance sorting by using p_i~*index is compared with the result of using clustering coefficient and degree index to sort node importance.The results show that the node importance ranking method based on the duplicato-degree and the neighborhood clustering coefficient is obviously better than the node importance ranking method based on the aggregation coefficient and degree.