In this talk, the problem of identifying the partial topology of complex dynamical networks via a pinning mechanism is presented. By using the network synchronization theory and the adaptive feedback controlling method, a method that can greatly reduce the number of nodes and observers in the response network are proposed. Particularly, this method can also identify the whole topology of complex networks. A theorem is established rigorously, from which some corollaries are also derived in order to make our method more cost-effective. In addition, an approach is also given to avoid possible identification failure caused by inner synchronization of the drive network.