Information visualization has attracted much attention in recent years in many fields of science and engineering. In many applications, graphs are `dynamic' in the sense that changes are constantly applied to a graph to reflect the evolution of the system behaviour represented by the graph. In the past, the concept of the so-called "mental map" has largely been ignored. Users often have to spend a lot of time relearning the redrawn graphs. This paper proposes an effective way to release the user from such kind of a distasteful job by maintaining a high degree of the "mental map" for general graphs when graphs are redrawn. Our experimental results suggest this new approach to be promising. .