Crowdbreaks: Tracking Health Trends using Public Social Media Data and Crowdsourcing

This repository contains all additional code/data/analysis for the paper "Crowdbreaks: Tracking Health Trends using Public Social Media Data and Crowdsourcing".

If you use the data or model please cite the work like this

Müller, Martin M., and Marcel Salathé. "Crowdbreaks: Tracking health trends using public social media data and crowdsourcing." Frontiers in public health 7 (2019).

or

@article{muller2019crowdbreaks,
  title={Crowdbreaks: Tracking health trends using public social media data and crowdsourcing},
  author={M{\"u}ller, Martin M and Salath{\'e}, Marcel},
  journal={Frontiers in public health},
  volume={7},
  year={2019},
  publisher={Frontiers Media SA}
}

Install

conda env create -f environment.yml

Download tweets

Generate a set of Twitter API keys and download the tweets using the following command:

python download_tweets.py -i ./data/vaccine_sentiment_data.csv -o ./data/tweets.jsonl --consumerkey XXX --consumersecret XXX --accesstoken XXX  --accesssecret XXX

Download vaccine sentiment model

wget https://crowdbreaks-public.s3.eu-central-1.amazonaws.com/models/fasttext_v1.ftz -o ./data/fasttext_v1.ftz