Upload
dallin-crosier
View
219
Download
1
Tags:
Embed Size (px)
Citation preview
Utilizing Social Media to Understand Human
Interaction with Extreme Media Events -
The Superstorm Sandy Beta Test
Arthur G. Cosby Somya D. Mohanty
National Weather ServiceOnline Webinar
Jul 16, 2013
• Social Networking and micro-blogging service
• Created in 2006
• 140 character tweets
• 140+ million users /400 million tweets per day
• Fast information propagation
• Our Access:• Real-time Firehose – Instantaneous acquisition of tweets• Historical Track – Tweets since 2006
Extreme Events and Social Media
• Traditional Methods• Telephone Survey• Invasive information acquisition
• Twitter• 170 million active users worldwide
• 48 million in U.S.
• ~26 million geo-located “human sensors”• Passive information collection
• Use Cases• Sandy Super-Storm• Moore Tornado
Tracking Tweets
• Geographic Bounding Boxes• Hurricane or Tornado path
• Keyword Searches• Complex searches on text within tweets
• User Tracking• Tracking any tweets either made by a user or
mentioning a user (i.e. @usNWSgov – National Weather Service twitter handle)
• Hashtag Tracking• Tracking on topics (i.e. #sandy)
Advantages of Tracking Social Media
• Network Resiliency• Mobile phone service is pretty resilient - in certain use
cases traffic doubled
• Real-time Visual Monitoring • Tracking of pictures posted of the event from twitter
users via Instagram, Vine, etc.
• Identification of Sub-events • Power Outages, Flooding, Disaster recovery
• Determine Human Mobility Patterns• Ability to help disaster recovery agencies assist
before, after and during and event
Advantages of Tracking Social Media
• Development of Predictive Algorithms• Utilizing historical data to create predictive models
capable of detecting future events• Predicting the extent of damages as a result of an
disaster
• Help and Assist Information Propagation • Developing organic networks in case of an event
need real-time information feedback.
• Prevent Incorrect Information Dissemination • Analyzing the information disseminated by the users
of the network for their validity in context to an event
Moore Tornado (OK)
• 138K geo-coded tweets – May 15th – May 30th
• Utilization• Structural analysis of buildings, roads and
infrastructure using posted pictures• Modeling predictive algorithms by extracting
parameters consistent with tweets from affected areas
• NSF Rapid Response Grant
Sandy SuperStorm
• 4.8M Tweets - Oct 27th – Nov 14th 2012
• Utilization• Real-time visual monitoring of posted pictures• Traffic Analysis for Resiliency• Sub-Event Analysis – Power Outage• Topic Analysis – Keyword and Hashtag Clouds• Trend Analysis – Occurrence of events relative to others• Sentiment Analysis – Feedback of public opinion
• U.S. Department of Health and Human Services • Office of the Assistant Secretary for Preparedness and
Response• Collaboration with New Jersey Mayors office and Harvard Law
School
Organic Help Networks
• Creating networks of help
• Offers to help
• Asking for help from organizations
• Asking for help from followers