The Research Of Nodes' Influence In Social Networks

Social networks describe the stable and complex relationships between people. Not only social networks can expand space of communicating, but also they change the platforms of information transmission from the reality to the rapidly developing Internet. Every node in social networks which is a real user plays a diverse role. The nodes have different influences in information dissemination. The research of node's information dissemination influence is related with public opinion guidance and merchandising in social networks, so it has both theoretical and practical significances. Certainly, the correlational studies become more and more popular in the field of social networks. This paper combines the characteristics of complex networks with social networks and regards the topology of network as a key to study the information dissemination model and the node's influence maximization problem. The contents of the research divide into the following two parts.In order to solve the problem that there is only one infected status of nodes according to the existing information dissemination models, this paper analyses the factors those are actually existing in social network and abstracts them into the model parameters. We also apply the dissemination theory in complex networks and take multiple infected statuses of nodes into account based on the classification of individual status in epidemic models. Finally, we introduce the human forgetting curve as a function to describe infected nodes' decay. By all these means, we propose a new information dissemination model in social networks. At last, we simulate this model in a real social network and compare the model with other information dissemination models. Simulation results show that the propagation process of the new model is more consistent with the information propagation trends in the real social networks than other models, variant parameters those are corresponding to the propagation factors cause the dissimilar spreading trends. The changing trend of dissemination speed and range are in conformity with the laws of information dissemination.Because greedy algorithms are not applicable to large-scale social networks, we adopt the thought of heuristic algorithms. This paper assesses the information spreading influence of a node by computing the adjacent and global influence. We solve the problem that the influence range of edge nodes is overlapping by removing the nodes and edges those are within the scope of seed node's influence area and updating the network. Finally, we propose a new influence maximization algorithm based on the independent cascade model. This paper simulates the algorithm in two social networks with different topology structures. The simulation results prove that the influence maximization algorithm based on removing range overlapping of the fringe nodes' influence can expand the range of nodes' influence, and the effect of the algorithm is better.