Skeleton of weighted social network

In the literature of social networks, understanding topological structure is an important scientific issue. In this paper, we construct a network from mobile phone call records and use the cumulative number of calls as a measure of the weight of a social tie. We extract skeletons from the weighted social network on the basis of the weights of ties, and we study their properties. We find that strong ties can support the skeleton in the network by studying the percolation characters. We explore the centrality of w -skeletons based on the correlation between some centrality measures and the skeleton index w of a vertex, and we find that the average centrality of a w -skeleton increases as w increases. We also study the cumulative degree distribution of the successive w -skeletons and find that as w increases, the w -skeleton tends to become more self-similar. Furthermore, fractal characteristics appear in higher w -skeletons. We also explore the global information diffusion efficiency of w -skeletons using simulations, from which we can see that the ties in the high w -skeletons play important roles in information diffusion. Identifying such a simple structure of a w -skeleton is a step forward toward understanding and representing the topological structure of weighted social networks.