Research On Immunization Technology Based On Random Walk With Nodes' Attributes

Complex network is a powerful tool for complex system modeling.How to guide and control the information dissemination and diffusion in complex network is a hot spot in the current complex networks research.The classical targeted immunization has achieved good results by sorting the degree centrality or betweenness centrality of network nodes.However,the degree centrality method is difficult to obtain the global information of the network in more cases,and the number of nodes in large-scale networks makes the high complexity betweenness centrality algorithm difficult to apply.The improvements based on the above methods are mostly aimed at the node degree value,and the effect is different.In this dissertation,combined with the idea of random walk,considering the influence of node topology and individual attributes on propagation,we use the form of multi-particle random walk to count and sort the traffic of network nodes,and propose random walk immune algorithm and improved algorithm for undirected network and directed weighted network respectively.The main work of this dissertation is as follows:(1)Based on undirected and unweighted networks,a multi particle random walk immune algorithm(RWI)is proposed.According to the walk form,the probability transfer matrix is deduced and the result is solved.Through the multi-particle random walk and process quantity statistics in the network nodes,not only the height value nodes can be identified,but also the bridge nodes between different communities can be identified well.The propagation experiments in different real networks also show that the proposed algorithm can effectively delay the propagation burst time,reduce the maximum infection rate when the propagation reaches the peak,and has low complexity.(2)In network propagation,the reasonable modeling of weighted network is helpful to the research of network immunity.In this dissertation,according to different scenes,combined with individual attributes and individual interaction strength,we extract attributes of real complex system,quantify weights and model directed network.Combined with the node attributes,the random walk immunization technology proposed can adapt to the changes of different network types.This dissertation uses different attribute quantization methods for the directed weighted network in different scenarios,and mainly takes the traffic network and social network as examples to apply the constructed network to the subsequent spread experiments.(3)Further analysis shows that the propagation of directed weighted network is affected by the node in-degree and out-degree.According to this restriction,the concept of node quantity and degree product is proposed.And the random walk immunization algorithm combined with the product is further improved.The results show that the improved method combined with the product can screen out the key nodes on the propagation path more efficiently,and the immune effect of the algorithm is significantly improved compared with the former.The work and experiments described in the dissertation expand the research ability of undirected network and directed network propagation immunization,in combination with random walk,attribute quantification and network modeling,it introduces new ideas and new methods for the study of network propagation dynamics.