Rank-Refining Seed Selection Methods for Budget Constrained Influence Maximisation in Multilayer Networks under Linear Threshold Model

A repository with the datasets, experimental environment, post-processing pipeline, and detailed results for the above project.

Authors: Michał Czuba, Piotr Bródka
Affiliation: Wrocław University of Science and Technology, Poland

Data

Networks used in experiments come from Piotr Bródka's repository with experiments on ICM model: https://github.com/pbrodka/SQ4MLN and from a website of Manilo De Domenico: https://manliodedomenico.com/data.php.

Configuration of the runtime

git submodule update --init
conda create --name ltm-seeding-mln python=3.10
conda activate ltm-seeding-mln
pip install -r submodules/network-diffusion/requirements/production.txt
pip install -e submodules/network-diffusion
pip install ipykernel seaborn pandas-profiling
python -m ipykernel install --user --name=ltm-seeding-mln

This repo works with git LFS, so please install it in order to pull large files!

Structure of the repository

.
├── README.md
├── _data_set                 -> networks used in experiments
├── _experiments
│   ├── examples              -> example configs to run experiments
│   ├── all_methods           -> raw results of evaluation of all methods on base nets
│   │   ├── all_results.csv   -> csv with aggregated and processed results
│   │   ...
│   │   └── <method name>     -> detailed logs for <method name>
│   └── top_methods           -> raw results of evaluation of top methods on large nets
├── _results                  -> generated results that were used in the paper
├── misc                      -> miscellaneous scripts to trigger experiments
├── runners                   -> scripts to execute experiments according to configs
├── run_experiments.py        -> main entrypoint ti trigger simulations
├── submodules                -> backbone library for simulations as a submodule
├── experiments.ipynb         -> doodles
├── postprocessing.ipynb      -> script to genreate `all_results.csv`
├── all_result_analysis.ipynb -> script to analyse results form base networks
├── top_result_analysis.ipynb -> script to analyse results from large networks
└── efficiency_maps.ipynb     -> script to obtain efficiency maps

Executing experiments

To run experiments with execute: run_experiments.py and provide proper CLI arguments, i.e. a path to configuration file and runner type. See examples in _experiments_/examples for inspirations.

Generating data visualisations

Please run postprocessing.ipynb to convert raw results. Then, depending to the stage of the experiments, execute all_result_analysis.ipynb or top_result_analysis.ipynb to obtain analysis from performed experiments.