This repository contains code for studying the network of fake news, including code and data for JF, NF, NK, and EB "Credibility in the News: Do we need to read?" WSDM 2018 workshop on Misinformation and Misbehavior online, http://jpfairbanks.com/mis2-2018.
This codebase includes both a python package and a collection of julia scripts for analyzing fake news. We use NLP tools in python including the fabulous SpaCy library, and classifiers in sklearn. The Belief Propagation that appears here is written in Julia using LightGraphs.jl graphs.
We use PostgreSQL and MongoDB to store structured and unstructured data regarding
the articles. For the purpose of reproducibility a static snapshot of the data
has been made available through the src/newgraphing/download.py script. In our
live system this analysis can be conducted against the data as it is updated
every 15 minutes.
Here are some instructions to get going with this project
- julia
- python, pip
- git-lfs
- Download the code
git clone github.com/jpfairbanks/newsgraphing && cd newsgraphing - Install all dependencies with
pip install . - Install all julia dependencies in the
src/fakeprop/REQUIREfile withPkg.add - Run the main script to generate figures
python ./main.py
If anything is unclear or doesn't work, let us know on the issues page. Feel free to ask for help by opening a new issue and we will take a look and answer your questions.
- James Fairbanks
- Nate Knauf
- Natalie Fitch
- David Ediger
- Erica Briscoe