This repository contains code to reproduce the results in the paper Detecting Coordinated Inauthentic Behavior in Likes on Social Media: Proof of Concept by Laura Jahn, Rasmus K. Rendsvig, and Jacob Stærk-Østergaard.
Please cite the paper when using the code:
Jahn, Laura, Rendsvig, Rasmus K., Stærk-Østergaard, Jacob, "Detecting Coordinated Inauthentic Behavior in Likes on Social Media: Proof of Concept", 2022.
@article{JahnRendsvig22Coordination,
author = {{Jahn, Laura and Rendsvig, Rasmus~K. and St\ae}rk-{\O}stergaard, Jacob}},
title = {{Detecting Coordinated Inauthentic Behavior in Likes on Social Media: Proof of Concept}},
year = {2022},
journal = {arXiv}
url = {\url{https://arxiv.org/abs/2305.07350}}
}
The ABM and the data analysis is done in R
. The following packages are loaded in the scripts:
parallel
, iterators
, doParallel
, RColorBrewer
, glmnet
, mclust
, DescTools
, cluster
, here
, tikzDevice
, plot.matrix
.
Parallel computing is done through forking.
This repository contains three folders:
-
/ABM_Classification: contains all scripts to run the ABM (generate vote data) and execute classification analysis on varying populations, as well as scripts that produce misclassification numbers and plots
-
/Data: here is where all data is stored that is produced in /ABM_Classification
-
/MCS: scripts to calculate majority correctness scores of Jury Selection Procedures
-
please see further READMEs throughout folder hierarchy for guidance.
** Change in variable naming **
You may encounter that we refer to our agents as genuine and nefarious in the code and further READMEs deeper in the folder hierarchy. Please note that in the process of this research project, we changed the name of genuine agents to authentic agents in the paper. This also affects a change in notation from G to A. Similarly, we changed the name of nefarious agents to inauthentic agents. Among the inauthentic agent types, formerly nefarious agent types, we renamed the Amplifier agents (A) to Booster agents (B). These changes do not have any consequences in the technical modelling steps.****
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 authors can be contacted at laurajahn [at] outlook [dot] de and rendsvig [at] gmail [dot] com.