Modeling telephone call networks with group structure from a complex network perspective

A large organization is composed of groups, which may represent departments, centers, administrative units, etc. In this paper, we analyze the internal call interactions among telephone numbers belonging to the same or different groups in a large organization. We observe that the node degree probabilities are almost the same for degrees less than 30 and they are found to obey a power-law when the degrees are large. Further, we find two typical cumulative probability distributions of the intergroup link weight for the group-nodes in the internal telephone group network and we also find that there is a limit to the group size. Based on the results, we propose a growth model using a combination of preferential and non-preferential attachments as well as group splitting. We show that the constructed network exhibits similar properties as the real-life internal telephone call/group network.