Analysis Of The Social Network’s Biggest Influence Based On The Overlapping Community Structure

Social networks influence maximization problem like "viral marketing" issue in the field of marketing, being under certain conditions, with certain rules to select a set of initial node, and then through the influence of the diffusion model, making the choice of the initial node can affect the maximum range. In the study, however, people often consider the network as the properties of the nodes and edges in the graph, without considering the community structure properties of the entire network, which leads to get experimental results cannot satisfy the complex structure of the reality of the network.In order to comprehensive consideration to the network community structure properties, and improve the accuracy of the influence maximization algorithm, in this article, this paper proposes a research method based on the maximization of overlapping community structure influence.First, this paper proposes a community clustering algorithm based on k- means algorithm. And k- means algorithm select the nodes randomly firstly and always choose the nearest node in the process of clustering. In this article both of the question were improved, the algorithm initial rough community structure.Secondly, the paper puts forward the improved algorithm of a community clustering algorithm. In the algorithm, the overlapping between nodes of the node to the club, in order to improve the clustering results, and overlapping communities can be found.Again, this paper proposes a initial node selection algorithm based on threshold maximum impact. In the algorithm, through judging a node even side to determine the potential impact of node, the largest potential influence choice every time the node as the initial node, and it is not a simple use of node degree to determine the influence of the node. Based on this algorithm,we propose a influence maximization algorithm based on overlapping community structure.Finally, select Slashdot Zoo social network data, and set up the experimental environment for validation. In the experiments, compared the k means clustering algorithm and improved algorithm; Contrast the impact scope of greedy algorithm and the influence of the proposed algorithm, and analyzes the experimental conclusion.