Given a large (e.g. social or conceptual) network, applying graph matching to identify common structures (=subgraphs) might give us new vocabulary to talk about such networks.