Research On Node Importance Measurement And Influence Blocking Maximization In Complex Networks

Node importance measurement and influence block maximization(IBM)problems are both key problems in complex network research.The former aims to measure the status of nodes in the network,and the latter aims to enable active nodes to block negative nodes to the greatest extent.Communication,both studies have important guiding significance to the fields of social sciences,economics,genetics,and physical sciences.But,the accuracy of node importance measurement and influence block maximization algorithm needs to be further improved,and some practical constraints should also be taken seriously.To this end,this paper proposes a new node influence measurement method,and carries out the study of influence block maximization based on time constraints.The main contents are as follows:Firstly,recognizing the limitations of individual node metrics,this paper combines structural hole indicators and degree center indicators to evaluate the value of nodes in proportion,and takes the value of the node's first-order neighbors into consideration to form a mixed metric indicator Mixedinf.Experiments show that the mixed index Mixedinf is more accurate in evaluating the value of nodes.Secondly,recognizing the limiting effect of time on the influence blocking process,this paper proposes the influence blocking maximization problem based on time limitation(TC-IBM),which aims to find a set of nodes within a limited time to maximize the blocked negative influence.In order to solve the TC-IBM problem,this paper further proposes a composite framework SDSS(Seed-Distribution and SeedSeletion),which contains two parts: seed allocation strategy and seed selection strategy.In addition,this paper proposes a novel seed selection strategy RMS(Reverse Multi Step walking),which can be nested in the SDSS framework.The high performance of SDSS framework and RMS algorithm was verified on five real networks.Finally,in order to reduce the negative impact of the influence overlap effect on the algorithm and improve the accuracy of the RMS algorithm,this paper conducts a deep screening of the set of candidate nodes of the RMS.This paper first uses the similarity threshold to initially screen the candidate nodes,and then uses the greedy strategy to accurately evaluate the few candidate nodes,so that the theoretical guarantee of the algorithm is more sufficient,and it is also very helpful to improve the effect of the algorithm.Experiments verify the improvement of algorithm performance.