Discovering Hidden Leaders through Email Log Based on Chance Discovery

Chance Discovery for making decisions is widely applied in complex real world. However, the complexity of the real world was sometimes beyond the reach of chance discovery of observable events. Some members in a community are dark, who are not frequently communicating to the others, but they create important impacts to the community, called hidden leaders. In this paper, we extend the Key Graph algorithm by the method of direct graph, called Key Graph_D, for analyzing the impact degree of members in the communication network. Furthermore, Key Graph_D evaluates the benefit of chance discovery and adaptively discovers the hidden leaders who issue orders as dark members. Analytically and illustrating with examples, we show that Key Graph_D achieves better performance of discovering hidden leaders than Key Graph.