Prepare

You need pip3 install networkx.

generate_graph.py: générer des graphs and save to files

The program will generate and draw 5 graphs with different values p and q (cf. tme3.pdf). The graph and its clusters are saved as files in /data.

label_propagation.py: Label Propagation algorithm

Run the program with python3 label_propagation.py <graph-filename>. The graph file could by benchmark_*.txt, amazon.txt or n400_*.txtin /data. The result(clusters) will be displayed and saved to /graph and its mapping in text will be saved to /result.

Louvain

See the README.md in /gen-louvain.

Divisive approach

Same usage as label_propagation.py. The program is implemented with divisive approach and it only works on graph data/benchmark_n128_e1024.txt as its complexity is O(n^2).

Benchmark

See the README.md in /LFR-Benchmark. We use:

./benchmark -N 128 -k 16 -maxk 16 -muw 0.1 -minc 32 -maxc 32 -beta 1
./benchmark -N 1000 -k 16 -maxk 16 -muw 0.1 -minc 32 -maxc 32 -beta 1

to generate the 2 graphs benchmark*.txt in /data.