This repository contains Jupyter notebooks that replicate the main results of the following paper:

Bhadani, S., Yamaya, S., Flammini, A. et al. Political audience diversity
and news reliability in algorithmic ranking. _Nat Hum Behav_ (2022).
https://doi.org/10.1038/s41562-021-01276-5

The following BibTex code can be imported into a citation manager for reference:

@Article{bhadani2022political,
  author   = {Bhadani, Saumya and Yamaya, Shun and Flammini, Alessandro and Menczer, Filippo and Ciampaglia, Giovanni Luca and Nyhan, Brendan},
  journal  = {Nature Human Behaviour},
  title    = {Political audience diversity and news reliability in algorithmic ranking},
  year     = {2022},
  issn     = {2397-3374},
  abstract = {Newsfeed algorithms frequently amplify misinformation and other low-quality content. How can social media platforms more effectively promote reliable information? Existing approaches are difficult to scale and vulnerable to manipulation. In this paper, we propose using the political diversity of a website’s audience as a quality signal. Using news source reliability ratings from domain experts and web browsing data from a diverse sample of 6,890 US residents, we first show that websites with more extreme and less politically diverse audiences have lower journalistic standards. We then incorporate audience diversity into a standard collaborative filtering framework and show that our improved algorithm increases the trustworthiness of websites suggested to users--especially those who most frequently consume misinformation--while keeping recommendations relevant. These findings suggest that partisan audience diversity is a valuable signal of higher journalistic standards that should be incorporated into algorithmic ranking decisions.},
  doi      = {10.1038/s41562-021-01276-5},
  url      = {https://doi.org/10.1038/s41562-021-01276-5},
}

The notebook was written in R version 3.6.2 and Python 3. A number of dependencies are to be installed before running it. A list of package requirements is provided in a separate file. The simplest method is to run a package manager such as pip or conda. It is recommended that all dependencies are installed in a separate environment.

The best way to do so is to use the Miniconda, which is a small bootstrap version of the largest Anaconda Python distribution (R is also available). Miniconda is available for most major operative systems (Windows, Linux, MacOS). Please note that installing R is not a prerequisite while using Miniconda. It will automatically install R and its essential libraries in the virtual environment through the environment_infodiversity.yml.

1. Download Miniconda from here: https://conda.io/miniconda.html
2. Follow the installation instruction for your platform: https://https://conda.io/projects/conda/en/latest/user-guide/install/index.html
3. Clone or download this repository

    git clone https://github.com/glciampaglia/InfoDiversity.git

4. Open a terminal and run:

    cd InfoDiversity
    conda env create --file environment_infodiversity.yml
    conda activate infodiversity_results

This should install all needed packages and activate the environment. And it will also download the data which is included in the data/ folder.

To open the notebooks, follow the instructions above for installing the Python environment and for obtaining the dataset files. From the location where you downloaded this repository launch Jupyter Notebook with:

    conda activate infodiversity_results
    jupyter notebook

This will open the notebook interface in your browser. It will show the contents of this repository. Click on the entry called Fig_1_to_3_R.ipynb to reproduce Figures 1 to 3 (this notebook is written in R) and Fig_4_to_6_suppl_python.ipynb to reproduce Figures 4 to 6 and all the supplementary figures (this notebook is written in Python).