We propose a new method for assessing agents influence in network structures, which takes into consideration nodes attributes, individual and group influences of nodes, and the intensity of interactions. This approach helps us to identify both explicit and hidden central elements which cannot be detected by classical centrality measures or other indices.