The original Leader Rank algorithm deals with finding leaders in Internet social networks, where the direction of an edge has a dissimilar meaning from our case:
If A points to (follows) B in Twitter, then B is considered to be a leader of A.
However, in our communication network, an edge from A to B suggests that A has sent B a message.
Therefore, assuming that a leader in a criminal network tends to be a sender rather than a receiver, each edge in G_(T_S ) has to be reversed to be compatible with Leader Rank’s design.
We denote by G_(T_S ) the reversed subnetwork induced by suspicious topics.