With the popularity of social networks, tremendous information flows are generated due to frequency social interactions. How information disseminates over social networks has received widespread attention from both industry and academics. Different from traditional communication ways, intelligent information diffusion over social networks exhibits one new feature, reinfection-reemergence, caused by social behaviors of individuals. That is, social individuals may be involved in the diffusion process of intelligent information many times. Meanwhile, there are two critical problems: individuals’ participation in many times and self-confirming behavior. Previous models have not taken into account these two problems. In this paper, a novel intelligent information diffusion phenomenon is modeled, which introduces a new “uncertain” psychological state into classic susceptible-infected-recovered model. In the diffusion process, the new proposed model characterizes a practical reinfection-reemergence scenario caused by the change of social attributes for individuals. To verify the dynamics properties, the proposed intelligent model is simulated in synthetic and real social network datasets. For the theoretical result, the outbreak threshold and the stability of equilibria are quantitatively addressed. As for the experiments, results indicate that intelligent information would tend to disappear or spread depending on the propagation threshold. Consequently, theoretical and simulation results can provide an insight on the controlling of intelligent information diffusion.