In order to study the influence maximization problem in complex networks, a heuristic framework was developed. Based on the in-depth analysis of information transmit process between node pairs, the probability of a node being in the informed state was obtained, and then an approximation of spreading influence of seed nodes was conducted through expectation calculation. A fast evaluation algorithm was proposed based on sequential seeding strategy. Specifically, simulation results on six real networks from various fields all show that the proposed algorithm is able to distinguish a small set of influential seed nodes. Moreover, the influence scope of the seed nodes selected by the method is significantly better than three benchmark influence maximization algorithms under three common measurements.