earticle

영어

The emerging of social networks opens opportunities for viral marketing. It is a fundamental issue to find a subset of diffusion nodes such that targeting them initially will maximize the range of the information spreading. The problem of finding the most influential spreaders is unfortunately NP-hard. The best known approximation algorithm has been proven to be with an approximation ratio of(1−1/e), however, the performance and the time complexity of the approximation algorithm are not suitable for large-scale social networks. In this paper, we propose a community-based diffuse efficiency algorithm, which is differing from approximation algorithm for mining diffusion nodes through the whole networks, our algorithm identity nodes from the view of the community. The algorithm encompass two steps: Firstly, detect the number of communities in the networks by taking into account information diffusion; and then a dynamic programming algorithm for selecting communities to find diffusion nodes. The performance of the proposed algorithms is evaluated by experiments on a data set of 4000 people call logs, the results show that the community based diffuse efficiency algorithm performs better than the other two algorithms.