Abstract: This paper adopts methodology of multi-agent based modeling to represent rumor spreading on small-world social network and to address impacts of rumor on pricing decision and profit of supplier. In the model, the process of rumor diffusion follows well-accepted ‘Susceptible-Infective-Removal’ (SIR) model. Simulation proved that mathematic analysis of SIR model overestimated the results of rumor diffusion while PageRank algorithm is a more ‘efficient’ method to choose the source node for rumor propagation according to all four proposed indicators. Sensitivity analysis of two parameters in SIR model, spreading rate and ignoring rate, indicates that the loss of profit will be less with the increasing of two parameters.

Key words: pricing, small world network, rumor spreading, multi-agent modeling and simulation