5. The Opportunity:
The Challenge:
Making decisions from a blob
Red Cross:
Where are
shelters? What
are road
conditions?
Farmer Brown:
Should I move my
chickens to higher
ground?
NOAA flood risk zone on Google Crisis Maps,
For South Carolina March 30th, 11:30pm
6. The Opportunity:
The Potential:
Science allows us to answer these questions
Hydrologist from University
of South Carolina: We
should at least identify the
floodplain
Social Scientists from
Charleston College:
Census data indicates
who is vulnerable!
NOAA flood risk zone on Google Crisis Maps,
For South Carolina March 30th, 11:30pm
13. Social Indicators of
Vulnerability
1. High % of young
children
2. High % of elderly
3. Poverty
4. High population
density
5. Low community
cohesion
Dr. Susan
Cutter, Social
Vulnerability Index
14.
15. EE API Code to Refine the Flood Risk
Our Socio-ecological Risk Model
Zone
Floods in Boulder, Colorado,
September 19th
Total Number of Counties in High
Risk Zone: 4
At risk County with most people:
Weld
State: Colorado
Number of people affected: 88,569
High
Medium
Low
52. ~1200 replies from Sandy aftermath alone
“How do I use the map? What does it mean?”
“How old or new is this information?”
“I see something different than what this map says.”
61. How should I share what's going on with my contacts outside the city? What is my employer advising us to do? What is
my school or teacher advising us to do? Will I lose my job if I leave? How much will it cost to evacuate? How do I move
my elderly or sick loved one? What is everyone else in my neighborhood doing? Are they evacuating? Where would I
evacuate to? What do city/parish officials advise? What do trusted local newscasters think? How does this storm
compare with others that I've experienced? How severe is it? What is the path of the storm? What is the status of
evacuation routes? What options are there if I have no car? How should I prepare my home for a potentially long-term
evacuation? What stores are still open for last-minute supplies? What supplies do I need to ride out the storm? What are
the essential things I need to take for a potentially long-term evacuation? Where can I get food and supplies along my
evacuation route? Where are all of my friends and family? How can we let everyone know where we ended up? And
check in with those who stayed behind? What are the essential things I need to know about my destination? How to
keep sick/elderly/young children safe on the road? Where can I get medical care along my evacuation route? Contact
information for emergency services where I ended up. What hotels and shelters have space? Who has gas? Where are
the open ATMs on my evacuation route? What is the wait time at stores and other service centers? Where are the
contraflow routes? Where are the backroads and alternate routes to avoid congestion? How to tap into charities and
donations to ease the cost of evacuation? How do I stay in touch with people if the phone are lines down? How can I
share the news about what's happening with the outside world? Who else in my neighborhood is also staying? Where
can I get news updates specific to my neighborhood? Where should I store my possessions? Are there are authorities in
my area sticking around (in case I need help)? What services are still up and running (e.g. police, fire)? If my home
floods, where is the nearest safe place? Is there anyone around that needs shelter? Where is everyone and what is their
status? Where can I get medical help? What pharmacies are open? How do I care for the sick or injured if there aren't
medical facilities near by? How can I alert the outside world to conditions in my area? What are the announcements by
city/local officials? What resources does my community have so we can share? How to alert others to resources I've
found? When will the national guard or police be here (for safety)? How safe are the conditions in my area? How do I
keep my area safe? When will power be back on in my area? Where in the area is there still power? Where is there WiFi?
My home is destroyed - where can I find shelter? Where can I get a hot shower? Where can I find ice (to keep food
fresh)? Where can I find clean water? What stores are open and what supplies do they have? Where can I get food?
How do I keep my food supply safe from spoilage? Where can I wash my clothes? Where can I get cleaning supplies?
Where can I get coupons for food, basic necessities? What is the wait time at stores and other service centers? Where
can I go to get donations (clothes, etc.)? What are the requirements and deadlines for financial aid? What kind of aid
can I qualify for? How can I dispute a rejection for aid? How can I keep track of where my applications stand? If I apply
for one type, will I be disqualified for another? I've lost important documents in the storm. What are the steps to building a
new life elsewhere? I need to find a new job or source of income since my job went with the storm. Where can I get
mental health or spiritual help for PTSD? What are the steps in rebuilding a damaged home? Are there new construction
regulations? Where can I get home repair supplies? What health issues should I watch out for (bad water, chemicals in
home)? How to deal with toxins and mold in my home? Who in community has working resources (washing
machine, car) to share? How to get the city to resume services, e.g. garbage pick-up? If services such as schools remain
closed, where do I send my kids? How can I lobby for services such as libraries and schools to be rebuilt? Where can we
80. O p e n n n wo v a t i o n
Sku Ikn orks
flickr/cattoo
81. Open Innovation
Drawing on the edges
See risk and high levels of failure as
inevitable and better shared
Reduced cost of conducting research and
development)
flickr/cattoo
128. • 9.1% of gov. expenditure on Education
• 20% Illiterate Man.
• 38% Illiterate Woman.
• The Background Picture Are a Black board and
desks!
Education Without Borders
129.
130.
131. Challenges:
• Introducing Crowdsourcing Model.
• Internet Access
• Mapping Skills
• Verification for inaccessible areas.
• Engaging Actors with the platform.
Education Without Borders
175. Elections monitoring
Methodology
a.) Planning and training
b.) Data collection and editing (OSM)
c.) Mapping
- Boundaries
- Polling stations
- Police posts/stations
- Emergency Humanitarian centers
d.) Map printing and distribution
e.) SMS and video reporting.
f.) Offline engagement (wall paintings
etc)
176. Elections monitoring
Methodology
a.) Planning and training
b.) Data collection and editing (OSM)
c.) Mapping
- Boundaries
- Polling stations
- Police posts/stations
- Emergency Humanitarian centers
d.) Map printing and distribution
e.) SMS and video reporting.
f.) Offline engagement (wall paintings
etc)
177. Civic engagement
Map Kibera joined other organizations
to form the Kibera Civic Watch
Consortium which among other things;
a.) Organized the first Kibera
Parliamentary Candidates debate
and
b.) Interviewed all political candidates
on camera about their commitments
and promises.
178. Community engagement
a.) Regular community screening
forums where short films were
watched and debated by
participants.
b.) Feed back forums where data and
blog stories are shared offline to
invited community members and
leaders.
c.) Map printing and distribution to
organizations and agencies that
need them.
d.) Map painting on community walls
and open spaces.
179. Community engagement
a.) Regular community screening
forums where short films were
watched and debated by
participants.
b.) Feed back forums where data and
blog stories are shared offline to
invited community members and
leaders.
c.) Map printing and distribution to
organizations and agencies that
need them.
d.) Map painting on community walls
and open spaces.
180. Collaborations
Map Kibera collaborated with
the Uchaguzi elections
Monitoring project and
Ushahidi.
a.) Sharing workspaces
b.) Verifying information
together and working with
each other’s sources.
c.) Sharing data-bases of
Humanitarian intervention
institutions/agencies
181. Collaborations
Map Kibera collaborated with
the Uchaguzi elections
Monitoring project and
Ushahidi.
a.) Sharing workspaces
b.) Verifying information
together and working with
each other’s sources.
c.) Sharing data-bases of
Humanitarian intervention
institutions/agencies
182. Success stories
• Map Kibera’s Security maps used
by Police and Peace builders to
enhance security and promote peace
in the local neighborhoods.
Printed Maps were distributed to the
Police, The local administration and
the District Peace Committees in
both locations.
183. • Map Kibera successfully used its
methodology and tools to monitor the
2013 general elections in Kibera,
Mathare and partially in Mukuru slums
reducing likelihood of violence to almost
zero in these election hotspot locations.
184. • Since Map Kibera began focusing on
dangerous or ‘black spots’ and mapping
them in Kibera and Mathare, New police
posts have been constructed in
response.
185. • Kibera and Mathare slums are now open
more open and accessible than before.
Generated map data has made
development agencies and devolved
government committees enhance their
intervention approaches to minimize
duplication of projects and design more
relevant interventions.
186. • Several agencies, researchers and
organizations continue to approach Map
Kibera for more data on various
thematic subjects covered by Map
Kibera’s mapping.
187. “Your maps have saved my life,
You should do this allover the country!”
Kilimani Police Station OCPD on receiving a
copy of the new elections map in Kibera.
February 2013.
191. Goal: to sort tweets into different categories in real-time
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
192. How do you classify 200+
tweets/minute?
Option 1: process what you can
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
193. How do you classify 200+
tweets/minute?
Option 2: lists of keywords
“Disaster” or “Damage” or …
or “Bridge” but not “Game”
or … “Donation” and “Money”
but not … or “Missing
person” or “Missing people”
or “Missing” and “child” …
and “Tornado alert” and
“Tsunami alert” and … and
… or … and … but not …
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
195. Supervised Learning
• Label a small number of tweets
– Crowdsourcing
• Convert tweets to numerical vectors
– Feature extraction: not seen by the user
• Create a mathematical model of each class
– Statistical learning: not seen by the user
• Automatically categorize new elements
– Output to be used by maps, reports, etc.
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
196. Supervised learning for Twitter
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
197. Challenges
• Effectiveness and generality
– Classify accurately for each crisis
– Work well across crises
• Easy to use!
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
198. Free and open source software ∙ Implements supervised learning for disaster-related
tweets ∙ Easy to use through a web-based interface ∙ Available for beta-testing ∙ As most
systems it may perform poorly or not at all during the testing phase ∙ Free and open
source software ∙ Implements supervised learning for disaster-related tweets ∙ Easy to
use through a web-based interface ∙ Available for beta-testing ∙ As most systems it will
perform poorly or not at all during the testing phase ∙ Free and open source software ∙
Implements supervised learning for disaster-related tweets ∙ Easy to use through a webbased interface ∙ Available for beta-testing ∙ As most systems it will perform poorly or not
at all during the testing phase ∙ Free and open source software ∙ Implements supervised
learning for disaster-related tweets ∙ Easy to use through a web-based interface ∙
Available for beta-testing ∙ As most systems it will perform poorly or not at all during the
testing phase ∙ Free and open source software ∙ Implements supervised learning for
disaster-related tweets ∙ Easy to use through a web-based interface ∙ Available for betatesting ∙ As most systems it will perform poorly or not at all during the testing phase ∙
Free and open source software ∙ Implements supervised learning for disaster-related
tweets ∙ Easy to use through a web-based interface ∙ Available for beta-testing ∙ As most
systems it will perform poorly or not at all during the testing phase ∙ Free and open
source software ∙ Implements supervised learning for disaster-related tweets ∙ Easy to
use through a web-based interface ∙ Available for beta-testing ∙ As most systems it will
perform poorly or not at all during the testing phase ∙ Free and open source software ∙
Implements supervised learning for disaster-related tweets ∙ All work and no play makes
Jack a dull boy ∙ Easy to use through a web-based interface ∙ Available for beta-testing ∙
As most systems it might go bananas or stop working during the testing phase ∙ Free
and open source software ∙ Implements supervised learning for disaster-related tweets
200. Changes to keywords or geo region are stored for future reference.
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
201. Tagger: easily create
automatic classifiers for
your collection. Each
classifier has its own set
of categories.
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
202. 1
2
You can ask volunteers to
provide training labels 1
, or create them yourself 2
.
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
203. Clickers:
AIDR automatically creates a publicly-visible landing page for volunteers.
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
204. This is what volunteers see: one tweet after another, with a
series of options. Tweets to classify are selected by AIDR to
maximize accuracy gains.
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
205. Behind the scenes, AIDR learns words and how to use them to
classify tweets, e.g. “katyperry” => ~no, “secrecy” => ~yes, etc.
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
207. Output
• Export to .csv from the Collector
– For off-line analysis
• Live data feed in JSON format
– For creating crisis maps, reports, etc.
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
208. Example application: Crisis Tracker
AIDR: Artificial Intelligence for Disaster Response
http://aidr.qcri.org/
209. Free and open source software ∙ Implements supervised learning for disaster-related
tweets ∙ Easy to useto the AIDR team: Muhammad Imran, Jakob∙ As most
Thanks through a web-based interface ∙ Available for beta-testing
systems it may perform poorly or not at all during the testing phase ∙ Free and open
Rogstadious, Ji Lucas & Patrick Meier. Qatar Computing
source software ∙ Implements supervised learning for disaster-related tweets ∙ Easy to
use through a web-based interface ∙ Available for beta-testing ∙ As most systems it will
Research Institute.
perform poorly or not at all during the testing phase ∙ Free and open source software ∙
Implements supervised learning for disaster-related tweets ∙ Easy to use through a webbased interface ∙ Available for beta-testing ∙ As most systems it will perform poorly or not
at all during the testing phase ∙ Free and open source software ∙ Implements supervised
learning for disaster-related tweets ∙ Easy to use through a web-based interface ∙
Available for beta-testing ∙ As most systems it will perform poorly or not at all during the
testing phase ∙ Free and open source software ∙ Implements supervised learning for
disaster-related tweets ∙ Easy to use through a web-based interface ∙ Available for betatesting ∙ As most systems it will perform poorly or not at all during the testing phase ∙
Free and open source software ∙ Implements supervised learning for disaster-related
tweets ∙ Easy to use through a web-based interface ∙ Available for beta-testing ∙ As most
systems it will perform poorly or not at all during the testing phase ∙ Free and open
source software ∙ Implements supervised learning for disaster-related tweets ∙ Easy to
use through a web-based interface ∙ Available for beta-testing ∙ As most systems it will
perform poorly or not at all during the testing phase ∙ Free and open source software ∙
http://aidr.qcri.org/
Implements supervised learning for disaster-related tweets ∙ All work and no play makes
Jack a dull boy ∙ Easy to use through a web-based interface ∙ Available for beta-testing ∙
As most systems it might go bananas or stop working during the testing phase ∙ Free
and open source software ∙ Implements supervised learning for disaster-related tweets
226. VOLUNTEERS
A HUGE SHOUT OUT AND THANK YOU
PLEASE SEND AN EMAIL OR SMS TO AT LEAST ONE OF YOUR
VOLUNTERERS AND SAY THANK YOU FOR DOING WHAT YOU DO
AND FOR WHO YOU ARE!!
Earth engine was built for science, to do global analysis to ask new questions but it has a use one step further, and use apply science apply the science of disasters at this scale.(watch up tones) (ill give you an example of a collaborative research project)-but we could leverage this technology together with science, what does that mean for crisis mapping. I’m gonna show you how we have leveraged google earth engine to answer this question,
One solution- use science to refine (PRACTICE SAYING ONE SENTENCE PER SLIDE). This takes years of developing and days to run the model
Who would need to do this on the fly? Crisis mappers- mention who this technology exists todayGoogle earth engine can take terabytes of information and can analyzed scientific methods and algorithms on a speed and scale that’s never been done before.(im using this platform in my research).
We pull from both social and natural science to write that algortihm. I study hydrology, and bessie, who here with me today in the front row studies social science.
When we run this algorithm it gives us a real time risk surface anywhere in the US, pinpointing the location and number of people most affected.
Make the transition to next section
Whatif
Text or phone (making the science faster and to more people) (website, or an text message), or by a more prepared red cross.When scientists, programmers, and disaster reliefe experts come together…
Yale and thanks this is a proof concept that we’ve built but in order to achieve its full potential we came to nairobi to find looking for collaborators at this conference- disaster relief managers, better programers, to help us test the accuracy, improving the science and find better applications for this.
IFRC one
IFRC two
Or process later
The coolest areas of AI don’t exist in reality.
Some of you know why someone would call Katy Perry a Prism.
Some of you know why someone would call Katy Perry a Prism.
Some of you know why someone would call Katy Perry a Prism.
The ignite talks before me… Wow…. Their slides , Amazing….. Now you have me!!!!! A bit of background to why this presentation is not “all singing and dancing as the professional ones before”
Crazy never again!!!!
Volunteers clicking from Mali!
IF YOU CAN LIKE A PIC ON FACEBOOKYOU TOO CAN BE DIGITAL HUMANITARIAN VOLUNTEERDEMOCRATIZE DIGITAL ENGAGEMENT IN SUPPORT OF DISASTER RESPONSE----- Meeting Notes (4/2/13 16:10) -----STOP AND GOPAUSEPRONOUNCESYNC GESTURESARMS OPEN WIDEPALM UP, PALM DOWNHAND CHESTINDEX FINGERSQUINT
IF YOU CAN LIKE A PIC ON FACEBOOKYOU TOO CAN BE DIGITAL HUMANITARIAN VOLUNTEERDEMOCRATIZE DIGITAL ENGAGEMENT IN SUPPORT OF DISASTER RESPONSE----- Meeting Notes (4/2/13 16:10) -----STOP AND GOPAUSEPRONOUNCESYNC GESTURESARMS OPEN WIDEPALM UP, PALM DOWNHAND CHESTINDEX FINGERSQUINT
TEXT = TWITTER, SMS, FACEBOOK UPDATES
PAKISTAN
MORE SMS IN REGION
IF YOU CAN LIKE A PIC ON FACEBOOKYOU TOO CAN BE DIGITAL HUMANITARIAN VOLUNTEERDEMOCRATIZE DIGITAL ENGAGEMENT IN SUPPORT OF DISASTER RESPONSE----- Meeting Notes (4/2/13 16:10) -----STOP AND GOPAUSEPRONOUNCESYNC GESTURESARMS OPEN WIDEPALM UP, PALM DOWNHAND CHESTINDEX FINGERSQUINT
GOOD TO MENTION THAT WE HAVE GEO CLICKERS AS WELL COMING ONLINE THAT WILL INTEGRATE DIRECTLY WITH STORY MAP. This would have helped with backlog of geo locating
Yolanda Output
In the background you will see slides of the first part of SBTF’s activation with DHN members. SBTF led this phase.The slides tell the story which I or Patrick are more than happy to elaborate on during the next few days I want to talk about he most important part of the activation.NOW TO BE SERIOUS!VOLUNTEERS ARE NOT FREE.THEY TAKE TIME,SUPPORT, TRAININGYOU DO NOT WANT TO ABUSE THEM BY JUST MAKING A PICTURE FOR THE HELL OF ITSHOW THEM HOW THEY ARE CONTRIBUTING AND WHAT THEY ARE CONTRIBUTING TOWARDS.DURING THE LAST WEEK OR SO MANY OF US HERE HAVE BEEN WORKING RATHER HARD. OUR VOLUNTEERS HAVE BEEN WORKING JUST AS MUCH IF NOT MORETHIS WAS ONE REASON THAT IS NOT SPOKEN ABOUT WHY MICROMAPPERS WAS CREATED“”””””””””””MICROMAPPING---------------- MAKING CRISIS MAPPING EASIER!””””””””””””””””””””””