HumAID: Human-Annotated Disaster Incidents Data from Twitter

Description of the dataset


The HumAID Twitter dataset consists of several thousands of manually annotated tweets that has been collected during 19 major natural disaster events including earthquakes, hurricanes, wildfires, and floods, which happened from 2016 to 2019 across different parts of the World. The annotations in the provided datasets consists of following humanitarian categories. The dataset consists only english tweets and it is the largest dataset for crisis informatics so far.


Datasets collection details:

Data collection details including event name, event type, location, # of tweets, and data collection period
Event name Event type Data collection period Link Location Total collected tweets Total annotated tweets
2016 Ecuador Earthquake Earthquake 2016/04/17 -- 2016/12/25 https://en.wikipedia.org/wiki/2016_Ecuador_earthquake Ecuador 1,756,267 1,594
2016 Canada Wildfires Fire 2016/05/06 -- 2016/05/27 https://en.wikipedia.org/wiki/2016_Fort_McMurray_wildfire Canada 312,263 2,259
2016 Italy Earthquake Earthquake 2016/08/24 -- 2016/08/29 https://en.wikipedia.org/wiki/August_2016_Central_Italy_earthquake Italy 224,853 1,240
2016 Kaikoura Earthquake Earthquake 2016/09/01 -- 2016/11/22 https://en.wikipedia.org/wiki/2016_Kaikoura_earthquake New Zealand 318,256 2,217
2016 Hurricane Matthew Hurricane 2016/10/04 -- 2016/12/05 https://en.wikipedia.org/wiki/Hurricane_Matthew Venezuela, Colombia, Jamaica, Hispaniola, Puerto Rico, Cuba, Turks and Caicos Islands, The Bahamas, United States, Canada 1,939,251 1,659
2017 Sri Lanka Floods Flood 2017/05/31 -- 2017/07/03 https://en.wikipedia.org/wiki/2017_Sri_Lanka_floods Sri Lanka 40,967 575
2017 Hurricane Harvey Hurricane 2017/08/17 -- 2017/09/19 https://en.wikipedia.org/wiki/Hurricane_Harvey Windward Islands, Suriname, Guyana, Nicaragua, Honduras, Belize, Cayman Islands, Yucatán Peninsula, United States 6,384,625 9,164
2017 Hurricane Irma Hurricane 2017/09/06 -- 2017/09/21 https://en.wikipedia.org/wiki/Hurricane_Irma Cape Verde, Leeward Islands , Cuba and Puerto Rico, Turks and Caicos Islands, The Bahamas, United States 1,641,844 9,467
2017 Hurricane Maria Hurricane 2017/09/16 -- 017/10/02 https://en.wikipedia.org/wiki/Hurricane_Maria Lesser Antilles (especially Dominica and the U.S. Virgin Islands), Puerto Rico, Dominican Republic, Haiti, Turks and Caicos Islands, The Bahamas, United States 2,521,810 7,328
2017 Mexico Earthquake Earthquake 2017/09/20 -- 2017/10/06 https://en.wikipedia.org/wiki/2017_Puebla_earthquake Mexico 361,040 2,036
2018 Maryland Floods Flood 2018/05/28 -- 2018/06/07 https://en.wikipedia.org/wiki/2018_Maryland_flood United States 42,088 747
2018 Greece Wildfires Fire 2018/07/24 -- 2018/08/18 https://en.wikipedia.org/wiki/2018_Attica_wildfires Greece 180,179 1,526
2018 Kerala Floods Flood 2018/08/17 -- 2018/09/12 https://en.wikipedia.org/wiki/2018_Kerala_floods India 850,962 8,056
2018 Hurricane Florence Hurricane 2018/09/11 -- 2018/11/17 https://en.wikipedia.org/wiki/Hurricane_Florence West Africa, Cape Verde, Bermuda, United States,  Canada 659,840 6,359
2018 California Wildfires Fire 2018/11/10 -- 2018/12/07 https://en.wikipedia.org/wiki/Camp_Fire_(2018) USA 4,858,128 7,444
2019 Cyclone Idai Hurricane 2019/03/15 -- 2019/04/16 https://en.wikipedia.org/wiki/Cyclone_Idai Mozambique, Malawi, Madagascar, Zimbabwe 620,365 3,944
2019 Midwestern U.S. Floods Flood 2019/03/25 -- 2019/04/03 https://en.wikipedia.org/wiki/2019_Midwestern_U.S._floods United States 174,167 1,930
2019 Hurricane Dorian Hurricane 2019/08/30 -- 2019/09/02 https://en.wikipedia.org/wiki/Hurricane_Dorian Lesser Antilles, Puerto Rico, The Bahamas (especially the Abaco Islands and Grand Bahama), Eastern United States, Eastern Canada, southern Greenland, Iceland 1,849,311 7,660
2019 Pakistan Earthquake Earthquake 2019/09/24 -- 2019/09/26 https://en.wikipedia.org/wiki/2019_Kashmir_earthquake Pakistan 122,939 1,991
Total 24,859,155 77,196

Downloads: Labeled data and other resources

We release our labeled data by maintaining the Twitter terms of service: https://developer.twitter.com/en/developer-terms/agreement-and-policy, which states that "all developers may provide up to 50,000 public Tweets Objects". Here we only release Twitter ID, text and label.

  • HumAID dataset version v1.0:
    • Event wise dataset (set1) 47868 tweets: download
    • Event wise dataset (set2) 29,328 tweets: Please fill out this form to receive this set
    • Event type dataset: download, consists of Tweet id and label. Text can be obtained from event-wise dataset.
    • All Combined dataset: download, consists of Tweet id and label. Text can be obtained from event-wise dataset.

Tweets downloader tool using tweet ids:

To download full tweets content from Twitter, you can use our Tweets downloader tool written in Java. The tool can make 180 API calls per 15 minutes, each API call allows to get up to 100 tweets i.e. it can download up to 72,000 tweets per hour:» Download Tweets downloader tool

Please cite the following papers, if you use any of these resources in your research.
  1. Firoj Alam, Umair Qazi, Muhammad Imran and Ferda Ofli, HumAID: Human-Annotated Disaster Incidents Data from Twitter, In ICWSM, 2021. [Bibtex]