? perl driver script 
=> wdijk
  trance_back_w();
  person's correlation

=>** compare dd/nn versus k-mean method

=> neighbour join tree on distance matrix
=> identify local clusters based on distance matrix
   cluster parition? using nj method, compare with hiarchival method

=> betweenness calculation

-> output shortest path distance matrix

-> UML of codes

-> memory leak, 
  change "new" to copy of objects. in Node?
  delete in main code part.

x enumerate all possible shortest-paths
  Woest case senerio for u,v, the (n-2) nodes for a fully connected component,
in which case, the possibility can be as high as n factorial.
  In a power-law network, this can be further reduced. 



--old tasks: 
x random sampling of nodes
x get Node* by input orfs. 
x randomize input nodes for random shortest path
