sraashis / graph-de-anonymization

Seed based and seed-free graph de-anonymization.

Implementation is available in the link Github Link Report is here

Install libraries:

• pip install -U numpy networkx tqdm torch pandas scikit-learn matplotlib

Final results are in the output folder

Project No. 1:

---

This is the implementation of graph de-anonymization in two ways:

• Seed Based
  • It includes iterative seed propagation technique by matching multi-hop neighborhood structure
• Seed Free
  • It includes multi-hop density matching to extract confident seed and use the previous seed propagation algorithm.

Usage for seed-based

- python seed_based_main.py -nw 32 -g1 seedbased/G1.edgelist -g2 seedbased/G2.edgelist -sm seedbased/seed_node_pairs.txt -out seedbased/seed_based_result.txt

  -nw NUM_WORKERS, --num_workers Number of workers/processes.
  -g1 G1_EDGELIST_FILE, --g1_edgelist_file, Path to g1 edgelist.
  -g2 G2_EDGELIST_FILE, --g2_edgelist_file, Path to g2 edgelist.
  -sm SEED_MAPPING_FILE, --seed_mapping_file, Path to g1->g2 seed nodes mapping.
  -out OUTPUT_FILE, --output_file,  Path to output file.
  -mpi MAP_PER_ITR, --map_per_itr, Number of nodes to map on each global iteration

Usage for seed-free

- python seed_free_main.py -nw 32 -g1 seedfree/G1.edgelist -g2 seedfree/G2.edgelist -out seedfree/seed_free_result.txt

 -nw NUM_WORKERS, --num_workers, Number of workers/processes.
 -g1 G1_EDGELIST_FILE, --g1_edgelist_file, Path to g1 edgelist.
 -g2 G2_EDGELIST_FILE, --g2_edgelist_file, Path to g2 edgelist.
 -sin SEED_INIT_NUM, --seed_init_num, Number of seed mappings to extract.
 -out OUTPUT_FILE, --output_file,  Path to output file.
 -mpi MAP_PER_ITR, --map_per_itr, Number of nodes to map on each global iteration

Full pipeline run

chmod x+u jobs.sh
./jobs.sh