You can find data to our projects below). The data we are sharing contain tweet IDs for three categories: "All", "place" (tweets which mention a place, which includes a geo bounding box), and "exact location" (tweets which contain latitude/longitude pairs). From the IDs the full tweets can be obtained using a script and xplanations provided here.
The data has been collected through the Twitter filter stream API using a list of keywords and languages (provided below). Note that in some cases the keywords/languages had to be updated after a while (e.g. as new terms emerged). Should you have requests for only a subset of the data (or any kind of aggregate data), please get in touch through firstname.lastname@example.org. IMPORTANT: Due to the high volume of messages collected through the COVID-19 stream, data collected after February 25 is incomplete and and will only contain a sampled subset of all messages posted on Twitter. If you plan to use this data for research purposes, please be aware of the potential biases this induces.
Thank you for giving attribution if you end up using the data.
Müller, Martin M., and Marcel Salathé. "Crowdbreaks: Tracking Health Trends Using Public Social Media Data and Crowdsourcing." Frontiers in public health 7 (2019).
COVID-19 disease outbreak
Keywords: wuhan, ncov, coronavirus, covid, sars-cov-2
Vaccine sentiment tracking
Keywords: vaccine, vaccination, vaxxer, vaxxed, vaccinated, vaccinating, vacine, overvaccinate, undervaccinate, unvaccinated