4. ➢Cannot trust social media or data from the public
○ Many official agencies use social media to push out real-time updates
○ Data from the public is often more timely and can be validated with images
➢Crowdsourcing is like social media monitoring and listening
○ Crowdsourcing means actively engaging a crowd to conduct a specific task
○ More than just public outreach, two-way feedback thru meaningful engagement
➢Official data is more accurate than crowdsourced data
○ Data from official channels often takes longer and not always accurate
○ Digital volunteers often have verification protocols and leverage official sources
Typical Misconceptions
6. September 10 - 20, 2018
Activation #3 - Hurricane Florence
10 Coordination Calls
26 Crowdsourcing Roll-Ups
15 Crowdsourcing Products from
8 Digital Volunteer Networks and
6 Private Sector Organizations
* Connect to Private Sector thru NBEOC
October 8 - 14, 2018
Activation #4 - Hurricane Michael
7 Coordination Calls
15 Crowdsourcing Products from
8 Digital Volunteer Networks and
6 Private Sector Organizations
11 Crowdsourcing & Social Listening Reports
* Waze & GasBuddy integrated in GIS Dashboards
October 6 - 20, 2017
Activation #1 - Hurricane Maria
5,700+ Volunteers from
6 Digital Volunteer Networks
17 Coordination Calls
10 Projects
2 Hackathons
Crowdsourcing Exchange at FEMA HQ
August 22 - 25, 2018
Activation #2 - Hurricane Lane
3 Coordination Calls
7 Crowdsourcing Products
6 Digital Volunteer Networks
* Official real-time updates posted on social media
* Monitored private sector and non-profit
crowdsourcing platforms (Waze, GasBuddy, RxOpen)
Crowdsourcing Unit Activations
7. 7
The Humanitarian OpenStreetMap Team (HOT) creates
and provides free, up-to-date maps with the help of
volunteer mappers during disasters around the world.
Volunteers: 5,400+, ~300 active daily
Start & End Date: 9/23 - Ongoing
Buildings mapped: 1.4 million buildings
Roads mapped: 45,000+ km
Puerto Rico Buildings & Roads Base Map
http://bit.ly/HOT-disasters2017
8. 8
Goal: Status of 71 Hospitals in Puerto Rico
Outcome: In 36 hours with 55+ volunteers,
SBTF updated 33 of 71 hospitals, whereas
officials only had 30 of 71 hospitals, and then
63 of 71 and 44 unlisted health facilities
using social media and other online channels to
curate official and unofficial sources.
This was used to build hospital resupply plans,
support volunteer medical groups, and prioritize
resources.
Standby Task Force (SBTF) was formed in 2010 to provide
volunteer online digital responses to humanitarian crises.
Hospital Status Map
http://bit.ly/SBTF-hospitals
11. ● Combined crowdsourcing products
with social listening input provided
by External Affairs into single report
organized by FEMA’s Lifelines
● Produced twice a day typically sent
around 12 Noon and 12 Midnight
● Reviewed and sent out by
Situational Awareness Section
Chiefs to all the staff at the National
Response Coordination Center and
FEMA Top Level Managers
Crowdsourcing and Social Listening Reports
Crowdsourcing Map Journal:
http://bit.ly/CrowdMichaelJournal
13. Partnered with GIS to
incorporate crowdsourced data
from GasBuddy and Waze
GIS Unit
● Integrate crowdsourced data into GIS dashboards
● Formalize crowdsource imagery analysis process
Integrated crowdsourcing
products into External Affairs
Social Listening Reports
External
Affairs
● Finalize “No Wrong Door” procedure
● Leverage EA tools to guide crowdsourcing efforts
Crowdsourcing Coordinators
deployed & embedded in NRCC
NRCC
● Establish Official Incident Support title
● Establish Crowdsourcing Unit in NRCC roster & seating
● Distribute crowdsourcing products & resources
Collaborated with NBEOC to
leverage private sector data
NBEOC
● Expand relationship with NBEOC and private sector
● Share crowdsourced projects regularly with NBEOC
Hosted Daily Coordination Calls &
Created Crowdsourcing Reports
Digital
Volunteer
Networks
● Recognize partnership with Digital Volunteer
Networks
● Grow network of crowdsourcing partners
AccomplishmentsPartners Next Steps and Opportunities
The main advantage of mining citizen reports through Twitter is speed. Rapid tweet-based earthquake detection can potentially fill the gap between the time when an earthquake occurs and the time when seismically derived information becomes available. TED detects two to three earthquakes a day, on average. Especially in regions with few seismometers, TED reports often come in before traditional seismic networks detect an earthquake, giving seismologists early warning. TED sometimes detects earthquakes entirely missed by USGS’s automatic processing system, thereby increasing the number of felt events known to the agency. In addition, the tweet text and attached images sometimes offer a rapid qualitative assessment of an earthquake’s impact.
http://earthquake.usgs.gov/earthquakes/ted
GIS- Hurricane Journal (I don’t think our crowdsourcing stuff was integrated into the Journal maybe in the Geospatial Call and some dashboards), CAP Imagery Analysis
EA, EMA- No wrong door process
NBEOC- formalize relationship between crowdsourcing and NBEOC, be able to leverage business data
Official IS title, DTS title, NRCC Roster, Seating
Recognition or Tasking Digital Volunteer Networks