This repository contains all code related to our term paper.
The following python packages are required to run the experiments:
- networkx
- matplotlib
- psutil
- memory_profiler
- sklearn
- scipy
- infomap
- numpy
They can all be installed in one command using pip: pip install -r requirements.txt
To generate a new LFR benchmark, simply modify the parameters in Parameters.py and LFR_gen.py and run the latter. The resulting files should appear in Graphs/LFR/....
The generated graphs can be previewed using Display_graph.py. The file can be set as a parameter atop the file. The graph drawing functionalities offered by NetworkX are not particularly strong, we recommend using external software such as Cytoscape (https://cytoscape.org/).
To run tests on saved networks, simply modify the parameters at the top of Run_test.py and run. The results should appear in the folders Found_communities and Results. The latter contains convenient CSV files.
The script Plotting.py can be used to plot the results from the generated CSV files.
The memory_profiler tool mprof.py can be used to measure memory usage: python mprof.py run <file>.py
And to plot the generated data: python mprof.py plot
Analysis.py contains code used to generate quantitative analysis of the results.
The functions used to analyze the graphs from snap can be directly downloaded and unpacked into a directory of choice. NB. Some graphs have overlapping communities.