When we consider the problem of finding influential nodes for information diffusion in a large-scale social network based on the
Independent Cascade Model (ICM)
, we need to compute the expected number of nodes influenced by a given set of nodes. However, a good estimate of this quantity needs a large amount of computation in the ICM. In this paper, we propose two natural special cases of the ICM such that a good estimate of this quantity can be efficiently computed. Using real large-scale social networks, we experimentally demonstrate that for extracting influential nodes, the proposed models can provide novel ranking methods that are different from the ICM, typical methods of social network analysis, and “PageRank” method. Moreover, we experimentally demonstrate that when the propagation probabilities through links are small, they can give good approximations to the ICM for finding sets of influential nodes.