关键词: 社会网络, 影响力最大化, 社区划分, 影响传播

Abstract: Influence maximization is a significant research direction in social networks. Its main purpose is to get the most influential users to make the range of influence diffusion maximizing. With the coming of big data, the traditional greedy algorithm can not overcome the time problem of influence maximization effectively because of high time complexity for large-scale social networks. This paper proposes the community division to solve influence maximization. Large-scale social networks are divided into smaller community modules using influence probability. Thus, the isolation between sub-communities is also eliminated at greatest degree considering the boundary nodes between communities. To improve the efficiency further, this paper considers an independent influence path as an influence evaluation unit in each community. At the same time, the most influential nodes are found utilizing parallel processing at every community. Finally, the paper verifies the feasibility and efficiency of the proposed algorithm by simulation experiment, which can adapt to the large-scale social networks better.

Key words: social network, influence maximization, community division, influence diffuse