Reproduction code for: How much research shared on Facebook is hidden from public view?

A comparison of public and private online activity around PLOS ONE papers

Authors: Asura Enkhbayar, Stefanie Haustein, Germana Barata, Juan Pablo Alperin

Resource	Link
Preprint	TBD
Article	TBD
Code	Zenodo (doi:10.5281/zenodo.3381821)
Data	Dataverse (doi:10.7910/DVN/3CS5ES)

---

This repository contains all figures and tables present in the manuscript for "How much research shared on Facebook is hidden from public view?". Output files can be found in:

• figures/ - contains all figures used in the manuscript
• tables/ - contains all programmatically created tables used in the manuscript

Furthermore, all the input data and code required to reproduce results are provided with instructions. Provided scripts include:

• download_data.sh - to download input data
• prepare_data.py - data preprocessing
• analysis.py - data analysis and outputs

This article is part of a broader investigation of the hidden engagement on Facebook. More information about the project can be found here.

Initial Data Collection

The data used in this paper was collected using our own methods. The data collection method is described in [Enkhbayar and Alperin (2018)(https://arxiv.org/abs/1809.01194)]. Code & instructions can be found here.

Reproduce results

All scripts have been written with Python 3.x. To explore results interactively a working instance of Jupyter Notebooks/Labs is required.

Packages specified in requirements.txt can be installed via

pip install -r requirements.txt

1. Clone this repository and cd into the scripts folder

   git clone git@github.com:ScholCommLab/fhe-plos-paper.git
   cd fhe-plos-paper/scripts
   

2. Download data from Dataverse.

   All the data is hosted on dataverse: Dataverse repository

   Using the helper script provided, you can download all files into the respective locations. Make the script executable and ensure that you have wget installed.

   chmod +x download_data.sh
   ./download_data.sh
   

3. Preprocess data

   Run the preprocessing script to apply transformations on the input dataset. This step creates the file data/articles.csv

   python process_data.py

4. (Re)produce results

   Run the analysis script to produce figures and tables.

   python analysis.py

   Optionally, you can also open the notebook analysis.ipynb with Jupyter to explore the dataset and results.