CrisisMMD: Multimodal Twitter Datasets from Natural Disasters

The CrisisMMD multimodal Twitter dataset consists of several thousand manually annotated tweets and images collected during seven major natural disasters, including earthquakes, hurricanes, wildfires, and floods from 2017. The dataset includes three types of annotations:

On HuggingFace, we hosted version 2.0 of the CrisisMMD dataset. Please see further information below.

Disaster Response Tasks

1. Task 1: Informative vs Not Informative

   • Informative
   • Not informative
   • "Don't know or can't judge" → Removed in version 2.0

2. Task 2: Humanitarian Categories

   • Affected individuals
   • Infrastructure and utility damage
   • Injured or dead people
   • Missing or found people
   • Rescue, volunteering, or donation effort
   • Vehicle damage
   • Other relevant information
   • "Not relevant or can't judge" → Updated to "Not humanitarian" in version 2.0

3. Task 3: Damage Severity Assessment

   • Severe damage
   • Mild damage
   • Little or no damage
   • "Don't know or can't judge"

Datasets Details

The keywords used for collecting tweets, along with the start and end dates for each event, are outlined in the following table.

Crisis Name	Keywords	Start Date	End Date
Hurricane Irma	Hurricane Irma, Irma storm, Storm Irma, etc.	Sep 6, 2017	Sep 21, 2017
Hurricane Harvey	Hurricane Harvey, Tornado, etc.	August 25, 2017	September 20, 2017
Hurricane Maria	Hurricane Maria, Maria Storm, etc.	September 20, 2017	November 13, 2017
California wildfires	California fire, USA Wildfire, etc.	October 10, 2017	October 27, 2017

Event-wise data distribution

For each event, we collected tweets and associated images, filtered and sampled them for the annotation.

**Data distribution from the CrisisMMD version v1.0**

Crisis Name	# Tweets	# Images	# Filtered Tweets	# Sampled Tweets	# Sampled Images
Hurricane Irma	3,517,280	176,972	5,739	4,041	4,525
Hurricane Harvey	6,664,349	321,435	19,967	4,000	4,443
Hurricane Maria	2,953,322	52,231	6,597	4,000	4,562
California wildfires	455,311	10,130	1,488	1,486	1,589
Mexico earthquake	383,341	7,111	1,241	1,239	1,382
Iraq-Iran earthquake	207,729	6,307	501	499	600
Sri Lanka floods	41,809	2,108	870	832	1,025
Total	14,223,141	576,294	36,403	16,097	18,126

Data preparation for multimodal baseline

For the multimodal baseline experiments, we first combined the tweet text and image from all events. It resulted in 24 duplicate entries (tweet ids: text and associated images). We manually checked these duplicate entries and kept the one, which were annotated properly. We changed the label “Not relevant or can’t judge” to “Not humanitarian”. In addition, as the annotation consists of a label - “don't know or can't not judge”, we also removed them for the classification experiments. Hence, this preprocessing part filtered out 39 tweets and associated 44 images. The resulted total dataset consists of 16058 and 18082 tweet texts and images, respectively as shown in the following table. This version of this dataset is released as version 2.0 and is available for download.

**Data distribution from the CrisisMMD version v2.0**

In this version, the "Not relevant or can't judge" label has been mapped to "Not humanitarian" for the humanitarian task. Additionally, the "Not informative" label from the informative task has also been mapped to "Not humanitarian" for the humanitarian task. Duplicate entries from different events have been removed.

Informativeness

	Text	Image
Informative	11,509	9,374
Not informative	4,549	8,708
Total	16,058	18,082

Humanitarian

	Text	Image
Affected individuals	472	562
Infrastructure and utility damage	1,210	3,624
Injured or dead people	486	110
Missing or found people	40	14
Not humanitarian	4,549	8,708
Other relevant information	5,954	2,529
Rescue, volunteering, or donation effort	3,293	2,231
Vehicle damage	54	304
Total	16,058	18,082

Damage Severity

	Text	Image
Little or no damage	-	475
Mild damage	-	839
Severe damage	-	2,212
Total	-	3,526

Downloads (Alternate options)

• CrisisMMD dataset version v2.0: Download labeled images and tweets (~1.8GB)
• Datasplit: Annotations Download
• Datasplit for multimodal baseline with agreed labels: Annotations Download

Citation

Please cite the following papers if you use any of these resources in your research.

1. Ferda Ofli, Firoj Alam, and Muhammad Imran, Analysis of Social Media Data using Multimodal Deep Learning for Disaster Response, In Proceedings of the 17th International Conference on Information Systems for Crisis Response and Management (ISCRAM), 2020, USA.

2. Firoj Alam, Ferda Ofli, and Muhammad Imran, CrisisMMD: Multimodal Twitter Datasets from Natural Disasters, In Proceedings of the 12th International AAAI Conference on Web and Social Media (ICWSM), 2018, Stanford, California, USA.

@InProceedings{crisismmd2018icwsm,
  author = {Alam, Firoj and Ofli, Ferda and Imran, Muhammad},
  title = {{CrisisMMD}: Multimodal Twitter Datasets from Natural Disasters},
  booktitle = {Proceedings of the 12th International AAAI Conference on Web and Social Media (ICWSM)},
  year = {2018},
  month = {June},
  date = {23-28},
  location = {USA}
}
@inproceedings{multimodalbaseline2020,
Author = {Ferda Ofli and Firoj Alam and Muhammad Imran},
Booktitle = {17th International Conference on Information Systems for Crisis Response and Management},
Keywords = {Multimodal deep learning, Multimedia content, Natural disasters, Crisis Computing, Social media},
Month = {May},
Organization = {ISCRAM},
Publisher = {ISCRAM},
Title = {Analysis of Social Media Data using Multimodal Deep Learning for Disaster Response},
Year = {2020}
}