"Natural communities"

Large scale structural analysis of networks often reveals beautiful pattern. But it is not always easy to interpret the results. At an expert meeting on “Diversity of research” a current project at the Humboldt University Frank Havemann gave a paper about so-called “natural communities”. The algorithm for clustering they propose starts from a certain strongly connected part of the network (a clique). The case they test method on is the field of Information Science. As a result one gets different “views” of groups or clusters on their intellectual environment. For some of these groups outside of their own is only one large “other world”. In other cases the “ego-perspective” is much more differentiated, and entails different groups. These individual perspectives – on the same network structure – can differ impressively. Depending where you stay (or better start) your perspective changes. I find that an example of an amazing differentiated picture drawn by computational analysis which keeps individual perspectives and still performs globally. See http://arxiv.org/abs/1008.1004

complex networks
natural communities