Research On Node Importance Analysis In Online Social Network

Node importance analysis is a research hotspot in the field of online social networks.Ranking influential nodes and identifying critical nodes are two of its important research contents.The former aims to rank the importance of all nodes in the network,the latter focuses on quickly and accurately identifying a set of critical nodes.For ranking influential nodes,the existing single-attribute ranking methods are not good enough for monotonicity and correctness.Recently,researchers proposed multi-attribute node ranking methods.However,most of these studies ignored the difference of contributions of different attributes,resulting in limited improvement in monotonicity and correctness.For identifying critical nodes,the existing methods mostly find this kind of nodes by network dismantling.However,they are not very targeted when deleting nodes during network decycling,causes many redundant nodes to be deleted.In addition,the centrality index used to evaluate the destructive ability of the node ignored the impact of connections between neighbor nodes.This paper presents a nodes importance ranking method based on node position and neighborhood.Firstly,this method improves the performance of evaluating the importance of the node location using the information of iterations of nodes in the K-core decomposition process.Then,the algorithm uses the importance of neighbor nodes to further improve the ability to distinguish the importance of nodes on the network edge.Finally,it adopts the entropy weight method to reasonably set the weight of position importance and neighbor importance.Experiment results show that the proposed method has high monotonicity,correctness and efficiency.This method can also maintain a good performance in different networks with different structure and provide more reasonable sorting results quickly and efficiently than other algorithms.This paper presents a method of identifying critical nodes based on the links of neighborhoods.In nodes deletion phase,this method designs a centrality index to evaluating the ability of the node to reduce the network connectivity by considering degree values of this node and its neighborhoods and the number of links between neighbor nodes.In the reinsertion phase,the proposed method designs a more reasonable strategy to put nodes back to the network for getting a smaller set of critical nodes.Compared with traditional methods,this method has no adjustable parameters and has stable performance in different scale and structure networks.It is more targeted to delete nodes,and more reasonable to reinsert nodes,and it can get a smaller set of critical nodes.