The knowledge diffusion process and social network structure are always evolving simutaneously. By taking into account the interaction frequency which is always used to measure the social closeness, the social knowledge diffusion (SKD) model for dynamic networks was presented. In the model, with probability p, the target node would preferentially select one neighbor node to transfer knowledge according to their interaction frequency instead of the knowledge distance. Otherwise, with probability 1-p, the target node would build a new link with one node in the system randomly. The simulation results show that, comparing with the random model defined by the random selection mechanism and the traditional knowledge diffusion (TKD) model driven by knowledge distance, the knowledge will spread more fast and more importantly, the network structure leads to an assortative one, which is a fundamental feature of social networks. The work is helpful for deeply understanding the coevolution of the knowledge diffusion and network structure.