Research On The Relationship Between Overlapping Area And Strong-weak Edges In Social Network

The rapid development of Internet promotes the progress of social information network.In the real world,information systems can be expressed by network structures,while network structures can be divided further into multiple sized communities,and the community is one of the most important characteristics for the community network.How to fast and effectively recognize the structure of network communities becomes an advanced research hotspot.Concerning that people always belong to multiple communities,the thesis researches on the recognition algorithm of overlapping communities,and on this basis,we further discuss the relationship between community structures of overlapping networks and strong-weak edges,and the main contributions of the thesis are as follows:(1)The traditional nonnegative matrix decomposition algorithm ignores the extensibility of network systems,which causes the high complexity of time and space in the process of the communities partition.Aiming at solving this problem,the thesis proposes an algorithm combined principal component analysis(PCA)with membership index(MI),and achieves the effective recognition of overlapping communities.Firstly,PCA technology is taken to dig key characteristics of networks;secondly,MI is taken to classify nodes.Experimental results show that this algorithm can fast and effectively detect overlapping community structures,and the module Q is close to the traditional algorithm.(2)Concerning the uniqueness of overlapping nodes in overlapping communities,and it is difficult to define the relationship between the overlapping and non-overlapping nodes by the traditional relationship between strong-weak edges.Aiming at this problem,the thesis proposes overlapping community coefficient index,and on this basis,we further re-definition the node degree and strong-weak edges by the weights,and the complex relationship hid in nodes is revealed;and on this basis,by changing the number of overlapping number of nodes,we research on the change of the number of strong-weak edges,and then the change of the number of strong-weak edges is analyzed based on the partition methods of strong-weak edges.(3)We research the impact of overlapping nodes and strong-weak edges on the information diffusion,and as a case study of infectious diseases,the thesis analyzes the diffusion velocity of infectious diseases in the four real-world data sets.Experimental results show that not only the external relationship but also the inherent relationship among nodes should be considered in the process of preventing infectious diseases,which can contribute to accurately find those important nodes and achieve the effective control of infection.The thesis explores the relationship of the community and strong-weak edges from the perspective of overlapping community structures.A novel node degree definition and the method for identifying strong-weak edges are proposed,which have the high theoretical meaning and practical value.