Node importance assessment is an important issue in complex network research, and it is important to study the vulnerability and robustness of complex networks. By considering the network topology and information dissemination, and quantifying the topology among local networks, a node importance evaluation LSC algorithm that considers the degree of nodes and the structural similarity of neighbor nodes was presented. Based on six classical real networks and two man-made networks, the network is attacked by means of intentional destruction. The experimental results of three evaluation indicators, i.e., maximum connectivity factor, network efficiency and network monotonicity, show that the proposed algorithm LSC can more accurately evaluate the importance of network nodes than other existing algorithms.