This repository contains code to reproduce the results in the paper Friction Interventions to Curb the Spread of Misinformation on Social Media by Laura Jahn, Rasmus K. Rendsvig, Alessandro Flammini, Filippo Menczer and Vincent Hendricks.
@article{Jahn23Friction,
author = {{Jahn, Laura and Rendsvig, Rasmus~K., and Flammini, Alessandro, and Menczer, Filippo, and Hendricks, Vincent}},
title = {{Friction Interventions to Curb the Spread of Misinformation on Social Media}},
year = {2023},
}
This work is based on the minimal social media simulation model SimSom: A Simulator for Social Media and based on the paper Vulnerabilities of the Online Public Square to Manipulation by Bao Tran Truong, Xiaodan Lou, Alessandro Flammini, and Filippo Menczer.
To reproduce the results of the paper on friction, scripts for data generation (networks), running the simulation, and data analysis can be found in Friction/
Networks are created before the simulation runs. The script to create networks Generate_networks.py is called through the script bash-nw.
Configs (parameter combinations) are generated with make_final_config_friction.py.
Our code is based on Python3.6+.
Activate virtualenv with the required packages and run pip install -e . for the module imports to work correctly.
We use conda, a package manager to manage the environments. Please make sure you have conda or mamba installed on your machine.
A snakemake file Snakefile controls the workflow and is called in final_res_bash.sh to start the simulation. Outputs are saved to Friction/results and Friction/verbose.
Installation times for packages and package managers are standard and shouldn't be a problem to a normal computer, and available to all standard operating systems. Runtime varies depending on parameters, and is faster when the simulation is deployed on multiple cores (as specified in bash and snakefiles).
Data is analysed in Analysis_results.ipynb.
The results in the paper are based on averages across multiple simulation runs. To reproduce those results, we suggest running the simulations in parallel, for example on a cluster. This decreases runtime significantly. Runtime increases with network size, number of runs, and convergence/stability parameters as described in the paper.
Run minimal example with workflow/example/run_simulation.py with the specified test follower network.
This project is licensed under the terms of the GNU General Public License v3.0 (gpl-3.0). See LICENSE for rights and limitations.
The corresponding author can be contacted at laurajahn [at] outlook [dot] de.