Classification of Temporal Graphs using Persistent Homology

Preliminaries

1. Make sure you have the requirements in requirements.txt installed. To do this, just run pip install -r requirements.txt in your python environment.
2. For the C++ part, make sure you have Eigen library installed. Then run the following commands in the terminal :

   g++ -o kmp kernelization.cpp -I/path/to/eigen -pthread -O3
   g++ -O3 -shared -std=c++17 -fPIC -o avg_all_diff.so avg_all_diff.cpp
   

Contact Network Models and Random Temporal Graphs

1. Add your graph classification parameters in the config.json file. Some sample parameters are added, with variable descriptions.
2. Run the script run_exp.py.

Real Datasets

1. The parse_dataset.py contains functions to parse files for highschool, hospital, mit, workplacev2. You can add your own function to parse a new data from a file.
2. run_exp_real.py will perform the classification task of RE vs CM or EWLS vs CM. The usage for this function is as follows :

   python run_exp_real.py dataset filepath exp_type num_perturb edge_swap time_swap diag_filter
   

   where

   positional arguments:
     dataset      Name of the dataset funtion in parse_datasets.
     filepath     Path to the dataset file.
     exp_type     Type of experiment : re or ewls.
     num_perturb  Number of re/ewls perturbations.
     edge_swap    Number of edge swaps for CM.
     time_swap    Number of time swaps for CM.
     diag_filter  Persistence threshold.
   

   For example,

   python run_exp_real.py hospital data/detailed_list_of_contacts_Hospital.dat_ re 25 50 50 0
   

   You can run python run_real_exp.py -h for help.