COVID-19 Twitter data, keyword stream 2020-01-13 to 2020-06-06

Twitter data was collected through the Twitter API, specifically through the filter streaming endpoint, using the Crowdbreaks platform (crowdbreaks.org) The data used in this work consists of a total of 353,993,900 tweets (thereof 267,026,740 retweets) posted by 26,262,332 users in a 146 day observation period, i.e. from January 13 to June 7, 2020. These tweets have been identified by Twitter to be in English language and match one or more of the keywords "wuhan", "ncov", "coronavirus", "covid" and "sars-cov-2".

The data is complete with respect to these keywords, except during a period between mid-March to mid-April when volume exceeded the 1% threshold imposed by Twitter and was subsampled by an (unknown) degree.

The following fields are published:

id: Tweet ID

is_retweet: Whether or not tweet is a retweet

num_retweets: Number of retweets

user.id: Id of tweeting user

country_code: country code as predicted by local-geocode (https://github.com/mar-muel/local-geocode)