Research On Nodes Importance Measurement In Social Networks

The Social network refers to a variety of complex social systems formed by social entities through social relations.It is the application of complex networks in the real world.Because of the complexity of social network structure,the diversity of characteristics and the difference of formation mode,the importance of nodes in social network will be different.Those special nodes that have a great impact on the structure,function or propagation process of the entire network are called important nodes.Once such nodes are found and destroyed,they will hinder the connection of the social network and even cause the social network paralysis.The nodes importance measurement in social networks is of great theoretical significance and practical value for the study of network robustness and vulnerability.For example,identifying important nodes can effectively control the spread of infectious diseases,can lock the opinion leaders,and improve marketing product efficiency.At present,the research on nodes importance measurement in social networks has become a common concern of many disciplines,which greatly promotes the development of interdisciplinary.In this paper,from the perspective of social network topology,the the nodes importance measurement is studied in parallel,and the accuracy of the proposed method is further evaluated by deleting nodes leading to the change of network connectivity.There are four main contents in this research as follows:Firstly,many methods of node importance measurement do not fully consider the "bridging" attribute of nodes,which leads to the one-sided problem of measurement results.This paper combines the idea of structural holes with the centrality feature to develop a local centrality measurement method for multi-level structural hole-oriented.This method considers the topological structure and scale of nodes and their first-order and second-order neighbors,and evaluates the nodes importance by using the direct and indirect constraints caused by the lack of first-level and second-level structural holes around the node.This method not only reflects the characteristics of local connection of nodes,but also can find important nodes in the social network when the global topology is unknown.It effectively solves the problem of high computational complexity of the global method,and is suitable for large-scale or non connected social network.Using this method to locate the important nodes in structural holes is of great significance to the application research of community discovery,information dissemination,and invulnerability to social networks oriented.Secondly,the importance of nodes in the social network is not only related to the local characteristics of the nodes and their neighbors,but also related to the location of the network where they are located.In this paper,the degree method and the shell decomposition method are combined organically,and a method of nodes importance measurement based on the multi-order neighbors' difference of kernel degree is proposed.On the one hand,this method makes up for the defect that the degree method only pays attention to the local information of the nodes in the network structure,but ignores its position information and multi-level neighbor environment information in the network,which causes the degree value to not fully represent the importance of nodes;On the other hand,it makes up for the defect of the shell decomposition method in coarse-grained analysis of node importance.At the same time,according to the average path value of the social network,the contribution of neighbor sets of different orders to the node importance is discussed,which well balances the accuracy of node importance measurement and the real network topology information.Experiments show that this method can measure the nodes importance more accurately than the classical methods and the computational complexity is low,which is suitable for the quantitative analysis of the nodes importance of large-scale network nodes.Thirdly,aiming at the limitation of using a single attribute to measure the nodes importance in social networks with different structures,this paper proposes a measurement method which integrates the location information of nodes,the size of neighboring nodes and the interaction force between nodes to evaluate the importance of nodes accurately and comprehensively.In this paper,firstly,the evaluation index of outward link diversity is proposed,which considers not only the network location of the node,but also the interaction with the neighbor nodes of different core layers.Theoretically,it makes up for the defects of k-shell decomposition method,such as coarse-grained division,only considering the residual degree,not applicable to BA model.Then,the two ideas of adjacency degree and structural hole are integrated,and the important nodes in the core position and structural hole position are considered.At last,the network simulation intentional attack experiment is carried out on the real data sets.It is found that the method proposed in this paper can better distinguish the differences between the important nodes,has more advantages in identifying the important nodes,and has stable performance,good measurement effect and strong robustness on the real networks with different topological structures.At last,aiming at the problem that it is difficult to obtain the whole structure of the large-scale real social network and the time complexity of defining the node importance by using the global information,a node importance measurement method based on the local attribute is proposed.In this paper,firstly,a local area network is constructed with the target node as the center and its neighbors,and the density of local area network is defined to express the degree of compactness between nodes.The density of local area network is easily affected by the scale of network and the nature of the relationship in the network,which makes it impossible to measure the difference of density between different networks objectively.Therefore,this paper introduces the adjacency degree and network assortativity coefficient which represent the scale of local area network and the connection tendency between nodes.This method comprehensively considers the scale,closeness and topological structure of the local area network where the node and its neighbors are located,and uses the influence of the local area network where the node is located in the entire network to evaluate the importance of the node.At last,considering the influence of network dynamics on the measurement results of the importance of nodes,a deliberate attack simulation experiment was conducted on the network in a static and dynamic manner on real data sets.The experiment proved that the method is suitable for large-scale and dynamically changing social networks.