Leveraging Facebook data for social good

Transcription

1 Leveraging Facebook data for social good Providing Value and Protecting Personal Information September 2018 Laura McGorman, Data for Good Paige Maas, Core Data Science

2 Safety Check Community Help Crisis Response

3 Early questions Can Facebook help fill critical data gaps during the first hours of a crisis? How do we transmit this information to the right people at the right time while preserving privacy?

4 Maps! Location: Where are people located during the crisis compared to where we expect them to be? Movement: How are people moving during the crisis? Connectivity: Where do people have access to network coverage? Power availability: Where are people able to charge their phones and access power? Safety Check: Where are people checking in safe? Where are they not? Displacement: Over the coming weeks and months are people able to return home?

5 Data Location records (latitude, longitude) People with the Facebook mobile app o who have opted in to location history o with connectivity o and phone turned on

6 Privacy Preserving Approach Aggregation, Smoothing, Filtering Aggregate into map tiles, generate summary statistics Remove tiles where the data is sparse

7 Privacy Preserving Approach Aggregation, Smoothing, Filtering Aggregate into map tiles, generate summary statistics Remove tiles where the data is sparse Aggregate into large (360K m^2) tiles Co-develop these resources with trusted partners Goal: to provide data-driven products that provide insights, not passively hand over data We share data in a privacy preserving way.

8 Maps! Location: Where are people located during the crisis compared to where we expect them to be? Typhoon Damrey, Vietnam

9 Maps! Movement: How are people moving during the crisis? Hurricane Harvey, Texas People leave Corpus Christi People travel inland to Dallas People shelter in place Movement begins to die off

10 Maps! Connectivity: Where do people have access to network coverage? Volcanic Eruption, Guatemala Before Volcanic Eruption After Volcanic Eruption

11 Maps! Displacement: Over the coming weeks and months are people able to return home? Napa Valley Fires, California

12 Methods Compare to Baseline How do crisis patterns look different from usual? Before During After

13 Hurricane Maria September 2017

14

15 FB, NetHope and Mapbox

16 FB and the American Red Cross

17 Kerala Floods August 2018

18 FB insights shared in near real-time with local humanitarian agencies

19 FB data led SEEDS India s response strategy

20 Guatemala Volcanic Eruption June 2018

21 Facebook Map on 2G Coverage after eruption Youth receiving information through FB messenger We want to know when we can go back to school, said Michelle, 15, in the Simon Bergaño Vi llegas shelter in Escuintla, so if you could send me this kind of information to my phone, that would be brilliant. This sentiment was echoed by Luis, 17, pictured, also in the Simon Bergañ o Villegas shelter, who signed up to U-Report as soon as we told him about the tool. We don t always know what is going on, so this seems like a great thing.

22 Results in the past year Partners: The Red Cross, the World Food Program, UNICEF, NetHope, Humanity Road, Direct Relief, SEEDS India, and others Number of disasters activated for: 100+, including hurricanes Harvey, Irma, and Maria, the California wildfires, Hurricane Florence, Typhoon Manghkut, and others.

23 How do I sign up? Attend this webinar good job! Confirm that you work o n natural disasters (not civil conflict, etc.) Sign data sharing agree ment if you re an NGO (longer, more creative process if you re a govt ) Figure out the user s within your org Get access to GeoInsights, join Slack Group Start Disaster Mapping!

24

25 Rules of the Road

26 Next Steps 1. Gabby if you d like to become a partner: gabriele.almon@nethope.org 2. Webinar #2 is on Oct 17 th please join us!

27 Thank you!