A High - Density Clustering Approach to Exploring the Structure of Social Networks

In this paper, we present a technique originally developed for decomposing complex software design problems, and suggest its usefulness for exploring the structure of large, non-directed social networks . The technique, based on a high-density clustering model defined on a graph, is quite efficient on very large, relatively sparse networks and provides a convenient, two-dimensional representation of the global network structure. In particular, we demonstrate the usefulness of the high-density clustering technique on the network defined by the interlocking directors of the 200 largest industrial corporations of the 1970 Fortune 800 . The technique enables us to examine the regions of "high-density interlocking" in this corporate subnetwork : regions where any group of firms is quite heavily interlock, and where any one firm in the group is not highly linked outside the group . These and other preliminary results indicate that the high-density clustering technique is conceptually appealing and requires much less time and computational expense than other exploratory methods employed .