The identifying of influential nodes in large-scale complex networks is an important issue in optimizing network structure and enhancing robustness of a system. To measure the role of nodes, classic methods can help identify influential nodes, but they have some limitations to social networks. Local metric is simple but it can only take into account the neighbor size, and the topological connections among the neighbors are neglected, so it can not reflect the interaction between the nodes. The global metrics is difficult to use in large social networks because of the high computational complexity. Meanwhile, in the classic methods, the unique community characteristics of the social networks are not considered. To make a trade off between affections and efficiency, a local structural centrality measure is proposed which is based on nodes' a nd their ‘neighbors’ structural holes. Both the node degree and “bridge” property are reflected in computing node constraint index. SIR (Susceptible-Infected-Recovered) model is used to evaluate the ability to spread nodes. Simulations of four real networks show that our method can rank the capability of spreading nodes more accurately than other metrics. This algorithm has strong robustness when the network is subjected to sybil attacks.