Assigning nodes to the nearest center

After the cluster centers (exemplars) are determined in the first phase of the algorithm, each node is assigned to the nearest center. Our implementation provides the following options:

Original method (follows the original AP implementations):
For every node we look for neighbors which are cluster centers and assign the node to the nearest one. If depth is greater than 1 then we allow indirectly connected nodes to be assigned to the center (this extends the original method).
Weighted BSF method (specific to our plugin):
In the case of incomplete graphs (especially sparse graphs), it is possible that none of the nodes neighbors will be a cluster center. When this option is selected, we look for shortest paths in the weighted network from nodes to cluster centers.