NIST (US National Institute of Standards and Technology – Data Science Symposium Proceedings
March 4-5, 2014
Abstracts for posters

[PDF] UTILIZING BIG SOCIAL MEDIA DATA FOR HUMANITARIAN ASSISTANCE AND DISASTER RELIEF
FRED MORSTATTER, SHAMANTH KUMAR, HUAN LIU
ARIZONA STATE UNIVERSITY

Disaster response agencies have started to incorporate social media as a source of fast-breaking information to understand the needs of people affected by the many crises that occur around the world. Twitter, one example of social media, produces over 500 million status updates each day. The volume of this data is too much for first responders to consume through manual analysis. New systems and approaches are needed to help first responders obtain situational awareness during a disaster using social media data. We discuss two social media tools that we have developed to assist first responders with the challenge of understanding this deluge of big social media data.

TweetTracker is a tweet collection and aggregation system that addresses the problem of collecting big social media data for first responders in disaster scenarios. Using TweetTracker, first responders enter queries related to a disaster as the event unfolds. From here, TweetTracker addresses the ETL process of our big data workflow. Once these queries are entered into the system, TweetTracker extracts matching tweets from Twitter using Twitter’s APIs. To answer queries in real-time, TweetTracker performs several transformations on each tweet, including: keyword extraction, user profile location translation, and several optimizations for indexing. Finally, TweetTracker

loads the TweetXplorer is a visualization system that addresses the challenge of understanding the big data generated during crisis on social media. It helps first responders to understand the data via some visual analytic components.

TweetXplorer focuses on emphasizing some key facets of disaster data to first responders, including: when relevant keywords are important, who are the most influential tweeters, and where are the geographic regions with the most requests for help.

These systems have helped first responders to find areas of need in recent crises including Typhoon Haiyan to generate an after-action report of the areas of need. They were also used during Hurricane Sandy to assess how social media could best be used to deliver aid. Our systems are the first of their kind to aid first responders in making sense of the fast, noisy, and big data generated on social media.