README authors: Bijaya Adhikari, Yao Zhang, and B. Aditya Prakash.
Dated: Januray 17, 2017. 


This is a reference implementation of the netCondense algorithm described in the following paper.

Condensing Temporal Networks using Propagation
Bijaya Adhikari, Yao Zhang, Aditya Bharadwaj, and B. Aditya Prakash
In 2017 SIAM International Conference on Data Mining (SDM 2017), April 2017, Houston, Texas.


Given a temporal network, netCondense returns a condensed temporal network with similar diffusive characteristics as the original network.
Please check the paper for more detail.

==========================================================================================================================================================================================

Requirement: Matlab


To see a Demo: 
   make demo
   
Please make sure that you have set the correct matlab path in makefile before execution. In my system, matlab path is /raid/R2014A/bin/matlab

Execute:
   python netCondense.py edgelist alpha_N alpha_T matlab_path

Example: python netCondense.py example\toy 0.9 0.5 matlab_path


==========================================================================================================================================================================================
Input:

edgelist:
time-stamp node_1 node_2 probability_1_to_2 probability_2_to_1 
1 1 2 0.1 0.2

alpha_N: a real number between 0 and 1 representing node reduction factor
alpha_T: a real number between 0 and 1 representing time reduction factor

================================================================================================================================================================================================
Output:

edgelist__final: the generated summary graph (edgelist with weights)

edgelist_superNodes: map for supernodes 10: 21, 32, 49, 35 means that nodes 10, 21, 32, 49, and 35 were merged and the super-node has id 10. 

edgelist_superTimes: map for supertoimes 1: 2, 3 means that time-stamps 1, 2, and 3 were merged and the super-time has id 1. 

edgelist_edgeScores: delta-scores for each edge in the union graph

edgelist_timeScores: delta-scores for each time-pairs


================================================================================================================================================================================================
Contact:

If you have any questions about the code, please contact Bijaya Adhikari at bijaya[at]cs[dot]vt[dot]edu 
