Research On Idetification Of Node Influence In Multi-layer Social Networks

With the continuous development of various advanced information technologies,a variety of social platforms have gradually developed.And the constant changes of the information's sources,dissemination modes,and content forms,also make the networks become more and more complex,so that the relevant theories of complex networks have attracted more and more attention.With the further exploration of research,a large number of scientific studies have found that the importance of nodes in social networks is different.The nodes with greater influence have important practical significance and wide application prospects.And they play important roles in controlling the spread of rumour,optimizing resource allocation,disseminating information efficiently,publishing advertisements accurately and so on.Based on this,this thesis will conduct research from the following aspects:Firstly,in view of the limitations of many current methods in identifying the different influences of nodes,this thesis is based on the k-shell method,introduces the weights of edges and influence coefficients of edges,and defines the concept of node's weighted degree in the single-layer networks.Based on this,the recognition algorithm of node influence in the single-layer social networks is proposed,which takes account of the nodes' own features,location features and local features.Secondly,the dissemination path of information has begun to evolve from single-layer social networks to multi-layer social networks.Therefore,based on the multi-layer network topology composed of multi-layer social networks in reality,this thesis improves the influence recognition algorithm in the single layer social networks of the previous text.Thereby,a method for identifying influential nodes in the multi-layer social networks is constructed,and the multi-layer weighted degree of each node is used to distinguish the different influences of the nodes.Finally,the influence recognition algorithm of single-layer networks is implemented in a representative Zachary karate club network,and the effectiveness of the single-layer algorithm is verified by comparing with other typical methods.At the same time,the influence recognition algorithm of multi-layer social networks are realized in the multi-layer networks of employees in the Department of Computer Science at Aarhus University,and the classical information diffusion model is utilized and improved to simulate the information dissemination process in the single-layer and multi-layer social networks.Thus,through comprehensive comparison,the multi-layer algorithm is proved to be accurate and effective.